Online Community Question Answering (CQA) platforms have become indispensable tools for users seeking expert solutions to their technical queries. The effectiveness of these platforms relies on their ability to identify and direct questions to the most knowledgeable users within the community, a process known as Expert Finding (EF). EF accuracy is crucial for increasing user engagement and the reliability of the provided answers. We present TUEF, a Topic-Oriented User-Interaction Model for EF , which aims to fully and transparently leverage the heterogeneous information available within online CQA platforms. TUEF integrates content and social data by constructing a multi-layer graph that maps user relationships based on their answering patterns on specific topics. By combining these sources of information, TUEF identifies the most relevant users for any given question and ranks them using learning-to-rank techniques. Our findings indicate that TUEF’s topic-oriented model significantly enhances performance, particularly in large communities discussing well-defined topics. Additionally, we show that the interpretable learning-to-rank algorithm integrated into TUEF offers transparency and explainability with minimal performance tradeoffs. The exhaustive experiments conducted across six CQA communities show that TUEF outperforms all competitors, achieving a minimum performance boost of 42.42% in P@1, 32.73% in NDCG@3, 21.76% in R@5, and 29.81% in MRR.

Recent advances in representation learning have enabled neural Information Retrieval (IR) systems to use learned dense representations for queries and documents to effectively handle semantics, language nuances, and vocabulary mismatch problems. In contrast to traditional IR systems that rely on word matching, dense IR models exploit query/document similarity in dense latent spaces to account for semantics. This requires substantial training data and comes with increased computational demands. Thus, it would be beneficial to predict how a system will perform for a given query to decide whether a dense IR model is the best option or alternatives should be used. Traditional Query Performance Prediction (QPP) models are designed for lexical IR approaches and perform sub-optimally when applied to dense neural IR systems. Therefore, there has been a renewed interest in QPP methods to improve their effectiveness for dense neural IR models. While the results of the new QPP methods are generally encouraging, there is ample room for improvement in absolute performance and stability. We argue that by using features more aligned with the underlying rationale of dense IR models, we can enhance the performance of QPP. In this respect, we propose the Projection-Displacement-Based QPP (PDQPP), which exploits the geometric properties of dense IR models, projects queries and retrieved documents onto subspaces defined by pseudo-relevant documents, and considers changes in retrieval scores within them as a proxy for retrieval coherence. Minor score changes suggest robust and coherent retrieval, while significant alterations indicate semantic divergence and potentially poor performance. Results over a wide range of experimental settings on both traditional (TREC Robust) and neural-oriented (TREC Deep Learning) test collections show that PDQPP mostly outperforms the state-of-the-art QPP baselines.

Dense Information Retrieval (IR) systems rely on neural networks to embed documents and queries within a latent low-dimensional space. Among the Dense IR approaches, bi-encoders are particularly popular, as they achieve state-of-the-art performance and allow for efficient encoding of documents and queries. Nevertheless, using this class of systems, by construction, all the documents and queries are represented using the same set of dimensions. In this article, we introduce the Manifold Clustering (MC) hypothesis which states that, for each query, there exists a query-dependent manifold of the original embedding space where the query and documents relevant to it cluster more effectively. We empirically validate the MC hypothesis showing that it is possible to find a query-dependent linear subspace of the original embedding space where high retrieval effectiveness is achieved. To find such subspaces, we propose the Dimension IMportance Estimators (DIMEs), a class of models that associate an importance score with each dimension of an embedding and can be used to project the dense representations only on the most important dimensions. We first demonstrate the effectiveness of the DIMEs by proposing an oracle DIME which employs annotated documents and induces performance improvements as big as +184% in terms of AP. To demonstrate the practical applicability of the DIMEs beyond the oracle, we also propose a set of DIMEs based on pseudo-relevance and active feedback that induce improvement as big as +49.6% in terms of AP and +55.9% in terms of nDCG@10. The effectiveness of such DIMEs not only empirically supports the MC hypothesis, but illustrates an actual strategy to outperform the state-of-the-art that does not require any form of retraining, fine-tuning or re-indexing and can be efficiently implemented at retrieval time.

We present a demo of RAGTrip, a modular conversational system that integrates Large Language Models (LLMs), spatial reasoning, and information retrieval to generate personalized walking itineraries in urban environments. Unlike traditional route planners or closed-book LLMs, RAGTrip interprets nuanced user preferences, avoids hallucinations, and grounds its suggestions in real-world geographic and factual data. The system features an interactive conversational interface that engages users in refining both the itinerary and the attractions to visit. Through dynamic map visualizations and contextual responses, users can explore and iteratively customize their routes. The demo includes a toggle to enable or disable Retrieval-Augmented Generation (RAG), allowing direct comparison between RAG-enhanced and closed-book LLM responses. This highlights the value of combining spatial and semantic grounding in conversational itinerary recommendation.

Conversational Information Access systems have experienced widespread diffusion thanks to the natural and effortless interactions they enable with the user. In particular, they represent an effective interaction interface for conversational search (CS) and conversational recommendation (CR) scenarios. Despite their commonalities, CR and CS systems are often devised, developed, and evaluated as isolated components. Integrating these two elements would allow for handling complex information access scenarios, such as exploring unfamiliar recommended product aspects, enabling richer dialogues, and improving user satisfaction. As of today, the scarce availability of integrated datasets - focused exclusively on either of the tasks - limits the possibilities for evaluating by-design integrated CS and CR systems. To address this gap, we propose CoSRec, the first dataset for joint Conversational Search and Recommendation (CSR) evaluation. The CoSRec test set includes 20 high-quality conversations, with human-made annotations for the quality of conversations, and manually crafted relevance judgments for products and documents. Additionally, we provide supplementary training data comprising partially annotated dialogues and raw conversations to support diverse learning paradigms. CoSRec is the first resource to model CR and CS tasks in a unified framework, enabling the training and evaluation of systems that must shift between answering queries and making suggestions dynamically.

Pre-trained language models based on transformer networks are highly effective for document re-ranking in ad-hoc search. Among these, cross-encoders stand out for their effectiveness, as they process query-document pairs through the entire transformer network to compute ranking scores. However, this traversal is computationally expensive. To address this, prior work has explored early-exit strategies, enabling the model to terminate the traversal of query-document pairs. These techniques rely on learned classifiers, placed after each transformer block, that decide if a query-document pair can be dropped. Diverging from previous approaches, we propose Similarity-based Early Exit (SEE), a novel-non-learned-strategy that exploits the similarities between query and document token embeddings to early-terminate the inference of documents that will most likely be non-relevant to the query. Even though SEE can be used after every transformer block, we show that the best advantage is achieved when applied before the first transformer block, thus saving most of the inference cost for the query-document pairs. Reproducible experiments on 17 public datasets covering in-domain and out-of-domain evaluation show that SEE can be effectively applied to four different cross-encoders, achieving speedups of up to 3.5× with a limited loss in ranking effectiveness.

Contextual dense representation models for text marked a shift in text processing, enabling a richer semantic understanding of the text and more effective Information Retrieval. These models project pieces of text into a latent space, describing them in terms of shared latent concepts, which are not explicitly tied to the text's content. Previous work has shown that certain dimensions of such dense text representations can be irrelevant and detrimental to retrieval effectiveness depending on the information need specified in the query. Higher effectiveness can be achieved by performing retrieval within a linear subspace that excludes these dimensions. Dimension IMportance Estimators (DIMEs) are models designed to identify such harmful dimensions, refining the representations of queries and documents to retain only the useful ones. Current DIMEs rely either on pseudo-relevance feedback, which often delivers inconsistent effectiveness, or on explicit relevance feedback, which is challenging to collect. Inspired by counterfactual modelling, we introduce Counterfactual DIMEs (CoDIMEs), designed to leverage noisy implicit feedback to assess the importance of each dimension. The CoDIME framework presented here approximates the relationship between a document's click frequency and its interaction with a given query dimension through a linear model. Empirical evaluations demonstrate that CoDIME outperforms traditional pseudo-relevance feedback-based DIMEs and surpasses other unsupervised counterfactual methods that utilize implicit feedback.

Pre-trained language models have been widely exploited to learn dense representations of documents and queries for information retrieval. While previous efforts have primarily focused on improving effectiveness and user satisfaction, response time remains a critical bottleneck of conversational search systems. To address this, we exploit the topical locality inherent in conversational queries, i.e., the tendency of queries within a conversation to focus on related topics. By leveraging query embedding similarities, we dynamically restrict the search space to semantically relevant document clusters, reducing computational complexity without compromising retrieval quality. We evaluate our approach on the TREC CAsT, 2019 and 2020 datasets using multiple embedding models and vector indexes, achieving improvements in processing speed of up to 10.3X with little loss in performance (4.3X without any loss). Our results show that the proposed system effectively handles complex, multi-turn queries with high precision and efficiency, offering a practical solution for real-time conversational search.

Academic Search is a search task aimed to manage and retrieve scientific documents like journal articles and conference papers. Personalization in this context meets individual researchers’ needs by leveraging, through user profiles, the user related information (e.g. documents authored by a researcher), to improve search effectiveness and to reduce the information overload. While citation graphs are a valuable means to support the outcome of recommender systems, their use in personalized academic search (with, e.g. nodes as papers and edges as citations) is still under-explored. Existing personalized models for academic search often struggle to fully capture users’ academic interests. To address this, we propose a two-step approach: first, training a neural language model for retrieval, then converting the academic graph into a knowledge graph and embedding it into a shared semantic space with the language model using translational embedding techniques. This allows user models to capture both explicit relationships and hidden structures in citation graphs and paper content. We evaluate our approach in four academic search domains, outperforming traditional graph-based and personalized models in three out of four, with up to a 10% improvement in MAP@100 over the second-best model. This highlights the potential of knowledge graph-based user models to enhance retrieval effectiveness.

Learning ranking models that are both explainable and effective is an emerging topic within the research area of explainable AI. Several Learning-to-Rank (LtR) algorithms have been recently proposed that build models that are simple to explain and, at the same time, almost as effective as their state-of-the-art, black-box counterparts. In this work, we propose Interpretable LambdaMART (ILMART), a novel framework with different strategies to constrain the state-of-the-art LtR LambdaMART algorithm to generate interpretable models, i.e., ensembles whose trees can use either single features (main effects) or a limited number of interacting features (interaction effects). ILMART facilitates a straightforward tradeoff between model explainability and effectiveness by precisely tuning the quantity of main and interaction effects during the learning phase. We show that slightly increasing their number allows ILMART models to reach ranking performances at par with full-complexity LambdaMART ones. Furthermore, reproducible experiments conducted on publicly available LtR datasets demonstrate that ILMART can improve nDCG@10 by up to 10% compared to state-of-the-art competitors while preserving an explainable structure. Finally, we explore the relationship between model explainability and inference efficiency by introducing a novel and easy-to-implement scoring algorithm for ILMART ranking models, achieving up to a \(100\times\) speedup compared to the baseline.

Learned dense representations are a popular family of techniques for encoding queries and documents using high-dimensional embeddings, which enable retrieval by performing approximate k nearest-neighbors search (A-kNN). A popular technique for making A-kNN search efficient is based on a two-level index, where the embeddings of documents are clustered offline and, at query processing, a fixed number N of clusters closest to the query is visited exhaustively to compute the result set. In this paper, we build upon state-of-the-art for early exit A-kNN and propose an unsupervised method based on the notion of patience, which can reach competitive effectiveness with large efficiency gains. Moreover, we discuss a cascade approach where we first identify queries that find their nearest neighbor within the closest t << N clusters, and then we decide how many more to visit based on our patience approach or other state-of-the-art strategies. Reproducible experiments employing state-of-the-art dense retrieval models and publicly available resources show that our techniques improve the A-kNN efficiency with up to 5x speedups while achieving negligible effectiveness losses. All the code used is available at https://github.com/francescobusolin/faiss_pEE

Recent advances in Information Retrieval have shown the effectiveness of embedding queries and documents in a latent high-dimensional space to compute their similarity. While operating on such high-dimensional spaces is effective, in this paper, we hypothesize that we can improve the retrieval performance by adequately moving to a query-dependent subspace. More in detail, we formulate the Manifold Clustering (MC) Hypothesis: projecting queries and documents onto a subspace of the original representation space can improve retrieval effectiveness. To empirically validate our hypothesis, we define a novel class of Dimension IMportance Estimators (DIME). Such models aim to determine how much each dimension of a high-dimensional representation contributes to the quality of the final ranking and provide an empirical method to select a subset of dimensions where to project the query and the documents. To support our hypothesis, we propose an oracle DIME, capable of effectively selecting dimensions and almost doubling the retrieval performance. To show the practical applicability of our approach, we then propose a set of DIMEs that do not require any oracular piece of information to estimate the importance of dimensions. These estimators allow us to carry out a dimensionality selection that enables performance improvements of up to +11.5% (moving from 0.675 to 0.752 nDCG@10) compared to the baseline methods using all dimensions. Finally, we show that, with simple and realistic active feedback, such as the user's interaction with a single relevant document, we can design a highly effective DIME, allowing us to outperform the baseline by up to +0.224 nDCG@10 points (+58.6%, moving from 0.384 to 0.608).

Rapid response, namely, low latency, is fundamental in search applications; it is particularly so in interactive search sessions, such as those encountered in conversational settings. An observation with a potential to reduce latency asserts that conversational queries exhibit a temporal locality in the lists of documents retrieved. Motivated by this observation, we propose and evaluate a client-side document embedding cache, improving the responsiveness of conversational search systems. By leveraging state-of-the-art dense retrieval models to abstract document and query semantics, we cache the embeddings of documents retrieved for a topic introduced in the conversation, as they are likely relevant to successive queries. Our document embedding cache implements an efficient metric index, answering nearest-neighbor similarity queries by estimating the approximate result sets returned. We demonstrate the efficiency achieved using our cache via reproducible experiments based on Text Retrieval Conference Conversational Assistant Track datasets, achieving a hit rate of up to 75% without degrading answer quality. Our achieved high cache hit rates significantly improve the responsiveness of conversational systems while likewise reducing the number of queries managed on the search back-end.

Understanding the behavior of deep neural networks for Information Retrieval (IR) is crucial to improve trust in these effective models. Current popular approaches to diagnose the predictions made by deep neural networks are mainly based on: i) the adherence of the retrieval model to some axiomatic property of the IR system, ii) the generation of free-text explanations, or iii) feature importance attributions. In this work, we propose a novel approach that analyzes the changes of document and query embeddings in the latent space and that might explain the inner workings of IR large pre-trained language models. In particular, we focus on predicting query/document relevance, and we characterize the predictions by analyzing the topological arrangement of the embeddings in their latent space and their evolution while passing through the layers of the network. We show that there exists a link between the embedding adjustment and the predicted score, based on how tokens cluster in the embedding space. This novel approach, grounded in the query and document tokens interplay over the latent space, provides a new perspective on neural ranker explanation and a promising strategy for improving the efficiency of the models and Query Performance Prediction (QPP).

Information Retrieval (IR) and Recommender Systems (RSs) tasks are moving from computing a ranking of final results based on a single metric to multi-objective problems. Solving these problems leads to a set of Pareto-optimal solutions, known as Pareto frontier, in which no objective can be further improved without hurting the others. In principle, all the points on the Pareto frontier are potential candidates to represent the best model selected with respect to the combination of two, or more, metrics. To our knowledge, there are no well-recognized strategies to decide which point should be selected on the frontier in IR and RSs. In this paper, we propose a novel, post-hoc, theoretically-justified technique, named "Population Distance from Utopia" (PDU), to identify and select the one-best Pareto-optimal solution. PDU considers fine-grained utopia points, and measures how far each point is from its utopia point, allowing to select solutions tailored to user preferences, a novel feature we call "calibration". We compare PDU against state-of-the-art strategies through extensive experiments on tasks from both IR and RS, showing that PDU combined with calibration notably impacts the solution selection.

Columnar data formats, such as Apache Parquet, are increasingly popular nowadays for scalable data storage and querying data lakes, due to compressed storage and efficient data access via data skipping. However, when applied to spatial or spatio-temporal data, advanced solutions are required to go beyond pruning over single attributes and towards multidimensional pruning. Even though there exist solutions for geospatial data, such as GeoParquet and SpatialParquet, they fall short when applied to trajectory data (sequences of spatio-temporal positions). In this paper, we propose TrajParquet, a format for columnar storage of trajectory data, which is highly efficient and scalable. Also, we present a query processing algorithm that supports spatio-temporal range queries over TrajParquet. We evaluate TrajParquet using real-world data sets and in comparison with extensions of GeoParquet and SpatialParquet, suitable for handling spatio-temporal data.

Thanks to recent advances in IR and NLP, the way users interact with search engines is evolving rapidly, with multi-turn conversations replacing traditional one-shot textual queries. Given its interactive nature, Conversational Search (CS) is one of the scenarios that can benefit the most from Query Performance Prediction (QPP) techniques. QPP for the CS domain is a relatively new field and lacks proper framing. In this study, we address this gap by proposing a framework for the application of QPP in the CS domain and use it to evaluate the performance of predictors. We characterize what it means to predict the performance in the CS scenario, where information needs are not independent queries but a series of closely related utterances. We identify three main ways to use QPP models in the CS domain: as a diagnostic tool, as a way to adjust the system's behaviour during a conversation, or as a way to predict the system's performance on the next utterance. Due to the lack of established evaluation procedures for QPP in the CS domain, we propose a protocol to evaluate QPPs for each of the use cases. Additionally, we introduce a set of spatial-based QPP models designed to work the best in the conversational search domain, where dense neural retrieval models are the most common approaches and query cutoffs are typically small. We show how the proposed QPP approaches improve significantly the predictive performance over the state-of-the-art in different scenarios and collections.

Neural approaches that use pre-trained language models are effective at various ranking tasks, such as question answering and ad-hoc document ranking. However, their effectiveness compared to feature-based Learning-to-Rank (LtR) methods has not yet been well-established. A major reason for this is because present LtR benchmarks that contain query-document feature vectors do not contain the raw query and document text needed for neural models. On the other hand, the benchmarks often used for evaluating neural models, e.g., MS MARCO, TREC Robust, etc., provide text but do not provide query-document feature vectors. In this paper, we present Istella22, a new dataset that enables such comparisons by providing both query/document text and strong query-document feature vectors used by an industrial search engine. The dataset consists of a comprehensive corpus of 8.4M web documents, a collection of query-document pairs including 220 hand-crafted features, relevance judgments on a 5-graded scale, and a set of 2,198 textual queries used for testing purposes. Istella22 enables a fair evaluation of traditional learning-to-rank and transfer ranking techniques on the same data. LtR models exploit the feature-based representations of training samples while pre-trained transformer-based neural rankers can be evaluated on the corresponding textual content of queries and documents. Through preliminary experiments on Istella22, we find that neural re-ranking approaches lag behind LtR models in terms of effectiveness. However, LtR models identify the scores from neural models as strong signals.

Interpretable Learning to Rank (LtR) is an emerging field within the research area of explainable AI, aiming at developing intelligible and accurate predictive models. While most of the previous research efforts focus on creating post-hoc explanations, in this paper we investigate how to train effective and intrinsically-interpretable ranking models. Developing these models is particularly challenging and it also requires finding a trade-off between ranking quality and model complexity. State-of-the-art rankers, made of either large ensembles of trees or several neural layers, exploit in fact an unlimited number of feature interactions making them black boxes. Previous approaches on intrinsically-interpretable ranking models address this issue by avoiding interactions between features thus paying a significant performance drop with respect to full-complexity models. Conversely, ILMART, our novel and interpretable LtR solution based on LambdaMART, is able to train effective and intelligible models by exploiting a limited and controlled number of pairwise feature interactions. Exhaustive and reproducible experiments conducted on three publicly-available LtR datasets show that ILMART outperforms the current state-of-the-art solution for interpretable ranking of a large margin with a gain of nDCG of up to 8%.

The sheer increase in volume of RDF data demands efficient solutions for the triple indexing problem, that is devising a compressed data structure to compactly represent RDF triples by guaranteeing, at the same time, fast pattern matching operations. This problem lies at the heart of delivering good practical performance for the resolution of complex SPARQL queries on large RDF datasets. We propose a trie-based index layout to solve the problem and introduce two novel techniques to reduce its space of representation for improved effectiveness. The extensive experimental analysis reveals that our best space/time trade-off configuration substantially outperforms existing solutions at the state-of-the-art, by taking 30-60% less space and speeding up query execution by a factor of 2-81 times.

Modern search engine ranking pipelines are commonly based on large machine-learned ensembles of regression trees. We propose LEAR, a novel - learned - technique aimed to reduce the average number of trees traversed by documents to accumulate the scores, thus reducing the overall query response time. LEAR exploits a classifier that predicts whether a document can early exit the ensemble because it is unlikely to be ranked among the final top-k results. The early exit decision occurs at a sentinel point, i.e., after having evaluated a limited number of trees, and the partial scores are exploited to filter out non-promising documents. We evaluate LEAR by deploying it in a production-like setting, adopting a state-of-the-art algorithm for ensembles traversal. We provide a comprehensive experimental evaluation on two public datasets. The experiments show that LEAR has a significant impact on the efficiency of the query processing without hindering its ranking quality. In detail, on a first dataset, LEAR is able to achieve a speedup of 3x without any loss in [email protected], while on a second dataset the speedup is larger than 5x with a negligible [email protected] loss (< 0.05%).

Conversational agents are drawing a lot of attention in the information retrieval (IR) community also thanks to the advancements in language understanding enabled by large contextualized language models. IR researchers have long ago recognized the importance o fa sound evaluation of new approaches. Yet, the development of evaluation techniques for conversational search is still an underlooked problem. Currently, most evaluation approaches rely on procedures directly drawn from ad-hoc search evaluation, treating utterances in a conversation as independent events, as if they were just separate topics, instead of accounting for the conversation context. We overcome this issue by proposing a framework for defining evaluation measures that are aware of the conversation context and the utterance semantic dependencies. In particular, we model the conversations as Direct Acyclic Graphs (DAG), where self-explanatory utterances are root nodes, while anaphoric utterances are linked to sentences that contain their missing semantic information. Then,we propose a family of hierarchical dependence-aware aggregations of the evaluation metrics driven by the conversational graph. In our experiments, we show that utterances from the same conversation are 20% more correlated than utterances from different conversations. Thanks to the proposed framework, we are able to include such correlation in our aggregations, and be more accurate when determining which pairs of conversational systems are deemed significantly different.

The sheer increase in volume of RDF data demands efficient solutions for the triple indexing problem, that is to devise a compressed data structure to compactly represent RDF triples by guaranteeing, at the same time, fast pattern matching operations. This problem lies at the heart of delivering good practical performance for the resolution of complex SPARQL queries on large RDF datasets. In this work, we propose a trie-based index layout to solve the problem and introduce two novel techniques to reduce its space of representation for improved effectiveness. The extensive experimental analysis, conducted over a wide range of publicly available real-world datasets, reveals that our best space/time trade-off configuration substantially outperforms existing solutions at the state-of-the-art, by taking 30-60 percent less space and speeding up query execution by a factor of 2 - 81×.

We observe that in curated documents the distribution of the occurrences of salient terms, e.g., terms with a high Inverse Document Frequency, is not uniform, and such terms are primarily concentrated towards the beginning and the end of the document. Exploiting this observation, we propose a novel version of the classical BM25 weighting model, called BM25 Passage (BM25P), which scores query results by computing a linear combination of term statistics in the different portions of the document. We study a multiplicity of partitioning schemes of document content into passages and compute the collection-dependent weights associated with them on the basis of the distribution of occurrences of salient terms in documents. Moreover, we tune BM25P hyperparameters and investigate their impact on ad hoc document retrieval through fully reproducible experiments conducted using four publicly available datasets. Our findings demonstrate that our BM25P weighting model markedly and consistently outperforms BM25 in terms of effectiveness by up to 17.44% in NDCG@5 and 85% in NDCG@1, and up to 21% in MRR.

Search engine ranking pipelines are commonly based on large ensembles of machine-learned decision trees. The tight constraints on query response time recently motivated researchers to investigate algorithms to make faster the traversal of the additive ensemble or to early terminate the evaluation of documents that are unlikely to be ranked among the top-k. In this paper, we investigate the novel problem of query-level early exiting, aimed at deciding the profitability of early stopping the traversal of the ranking ensemble for all the candidate documents to be scored for a query, by simply returning a ranking based on the additive scores computed by a limited portion of the ensemble. Besides the obvious advantage on query latency and throughput, we address the possible positive impact on ranking effectiveness. To this end, we study the actual contribution of incremental portions of the tree ensemble to the ranking of the top-k documents scored for a given query. Our main finding is that queries exhibit different behaviors as scores are accumulated during the traversal of the ensemble and that query-level early stopping can remarkably improve ranking quality. We present a reproducible and comprehensive experimental evaluation, conducted on two public datasets, showing that query-level early exiting achieves an overall gain of up to 7.5% in terms of [email protected] with a speedup of the scoring process of up to 2.2x.

Deep pretrained transformer networks are effective at various ranking tasks, such as question answering and ad-hoc document ranking. However, their computational expenses deem them cost-prohibitive in practice. Our proposed approach, called PreTTR (Precomputing Transformer Term Representations), considerably reduces the query-time latency of deep transformer networks (up to a 42x speedup on web document ranking) making these networks more practical to use in a real-time ranking scenario. Specifically, we precompute part of the document term representations at indexing time (without a query), and merge them with the query representation at query time to compute the final ranking score. Due to the large size of the token representations, we also propose an effective approach to reduce the storage requirement by training a compression layer to match attention scores. Our compression technique reduces the storage required up to 95% and it can be applied without a substantial degradation in ranking performance.

In precision-oriented tasks like answer ranking, it is more important to rank many relevant answers highly than to retrieve all relevant answers. It follows that a good ranking strategy would be to learn how to identify the easiest correct answers first (i.e., assign a high ranking score to answers that have characteristics that usually indicate relevance, and a low ranking score to those with characteristics that do not), before incorporating more complex logic to handle difficult cases (e.g., semantic matching or reasoning). In this work, we apply this idea to the training of neural answer rankers using curriculum learning. We propose several heuristics to estimate the difficulty of a given training sample. We show that the proposed heuristics can be used to build a training curriculum that down-weights difficult samples early in the training process. As the training process progresses, our approach gradually shifts to weighting all samples equally, regardless of difficulty. We present a comprehensive evaluation of our proposed idea on three answer ranking datasets. Results show that our approach leads to superior performance of two leading neural ranking architectures, namely BERT and ConvKNRM, using both pointwise and pairwise losses. When applied to a BERT-based ranker, our method yields up to a 4% improvement in MRR and a 9% improvement in [email protected] (compared to the model trained without a curriculum). This results in models that can achieve comparable performance to more expensive state-of-the-art techniques.

The identification of relevance with little textual context is a primary challenge in passage retrieval. We address this problem with a representation-based ranking approach that: (1) explicitly models the importance of each term using a contextualized language model; (2) performs passage expansion by propagating the importance to similar terms; and (3) grounds the representations in the lexicon, making them interpretable. Passage representations can be pre-computed at index time to reduce query-time latency. We call our approach EPIC (Expansion via Prediction of Importance with Contextualization). We show that EPIC significantly outperforms prior importance-modeling and document expansion approaches. We also observe that the performance is additive with the current leading first-stage retrieval methods, further narrowing the gap between inexpensive and cost-prohibitive passage ranking approaches. Specifically, EPIC achieves a [email protected] of 0.304 on the MS-MARCO passage ranking dataset with 78ms average query latency on commodity hardware. We also find that the latency is further reduced to 68ms by pruning document representations, with virtually no difference in effectiveness.

In a conversational context, a user expresses her multi-faceted information need as a sequence of natural-language questions, i.e., utterances. Starting from a given topic, the conversation evolves through user utterances and system replies. The retrieval of documents relevant to a given utterance in a conversation is challenging due to ambiguity of natural language and to the difficulty of detecting possible topic shifts and semantic relationships among utterances. We adopt the 2019 TREC Conversational Assistant Track (CAsT) framework to experiment with a modular architecture performing: (i) topic-aware utterance rewriting, (ii) retrieval of candidate passages for the rewritten utterances, and (iii) neural-based re-ranking of candidate passages. We present a comprehensive experimental evaluation of the architecture assessed in terms of traditional IR metrics at small cutoffs. Experimental results show the effectiveness of our techniques that achieve an improvement of up to $0.28$ (+93%) for [email protected] and $0.19$ (+89.9%) for [email protected] w.r.t. the CAsT baseline.

We observe that most relevant terms in unstructured news articles are primarily concentrated towards the beginning and the end of the document. Exploiting this observation, we propose a novel version of the classical BM25 weighting model, called BM25 Passage (BM25P), which scores query results by computing a linear combination of term statistics in the different portions of news articles. Our experimentation, conducted using three publicly available news datasets, demonstrates that BM25P markedly outperforms BM25 in term of effectiveness by up to 17.44% in [email protected] and 85% in [email protected]

In this paper we explore the use of Continuation Methods and Curriculum Learning techniques in the area of Learning to Rank. The basic idea is to design the training process as a learning path across increasingly complex training instances and objective functions. We propose to instantiate continuation methods in Learning to Rank by changing the IR measure to optimize during training, and we present two different curriculum learning strategies to identify easy training examples. Experimental results show that simple continuation methods are more promising than curriculum learning ones since they allow for slightly improving the performance of state-of-the-art λ-MART models and provide a faster convergence speed.

Query Expansion (QE) techniques expand the user queries with additional terms, e.g., synonyms and acronyms, to enhance the system recall. State-of-the-art solutions employ machine learning methods to select the most suitable terms. However, most of them neglect the cost of processing the expanded queries, thus selecting effective, yet very expensive, terms. The goal of this paper is to enable QE in scenarios with tight time constraints proposing a QE framework based on structured queries and efficiency-aware term selection strategies. In particular, the proposed expansion selection strategies aim at capturing the efficiency and the effectiveness of the expansion candidates, as well as the dependencies among them. We evaluate our proposals by conducting an extensive experimental assessment on real-world search engine data and public TREC data. Results confirm that our approach leads to a remarkable efficiency improvement w.r.t. the state-of-the-art: a reduction of the retrieval time up to 30 times, with only a small loss of effectiveness.

Learning an effective ranking function from a large number of query-document examples is a challenging task. Indeed, training sets where queries are associated with a few relevant documents and a large number of irrelevant ones are required to model real scenarios of Web search production systems, where a query can possibly retrieve thousands of matching documents, but only a few of them are actually relevant. In this paper, we propose Selective Gradient Boosting (SelGB), an algorithm addressing the Learning-to-Rank task by focusing on those irrelevant documents that are most likely to be mis-ranked, thus severely hindering the quality of the learned model. SelGB exploits a novel technique minimizing the mis-ranking risk, i.e., the probability that two randomly drawn instances are ranked incorrectly, within a gradient boosting process that iteratively generates an additive ensemble of decision trees. Specifically, at every iteration and on a per query basis, SelGB selectively chooses among the training instances a small sample of negative examples enhancing the discriminative power of the learned model. Reproducible and comprehensive experiments conducted on a publicly available dataset show that SelGB exploits the diversity and variety of the negative examples selected to train tree ensembles that outperform models generated by state-of-the-art algorithms by achieving improvements of [email protected] up to 3.2%.

Learning-to-Rank (LtR) solutions are commonly used in large-scale information retrieval systems such as Web search engines, which have to return highly relevant documents in response to user query within fractions of seconds. The most effective LtR algorithms adopt a gradient boosting approach to build additive ensembles of weighted regression trees. Since the required ranking effectiveness is achieved with very large ensembles, the impact on response time and query throughput of these solutions is not negligible. In this article, we propose X-CLE a VER, an iterative meta-algorithm able to build more efficient and effective ranking ensembles. X-CLE a VER interleaves the iterations of a given gradient boosting learning algorithm with pruning and re-weighting phases. First, redundant trees are removed from the given ensemble, then the weights of the remaining trees are fine-tuned by optimizing the desired ranking quality metric. We propose and analyze several pruning strategies and we assess their benefits showing that interleaving pruning and re-weighting phases during learning is more effective than applying a single post-learning optimization step. Experiments conducted using two publicly available LtR datasets show that X-CLE a VER can be successfully exploited on top of several LtR algorithms as it is effective in optimizing the effectiveness of the learnt ensembles, thus obtaining more compact forests that hence are much more efficient at scoring time.

Large popular events are nowadays well reflected in social media fora (e.g. Twitter), where people discuss their interest in participating in the events. In this paper we propose to exploit the content of non-geotagged posts in social media to build machine-learned classifiers able to infer users' attendance of large events in three temporal periods: before, during and after an event. The categories of features used to train the classifier reflect four different dimensions of social media: textual, temporal, social, and multimedia content. We detail the approach followed to design the feature space and report on experiments conducted on two large music festivals in the UK, namely the VFestival and Creamfields events. Our attendance classifier attains very high accuracy with the highest result observed for the Creamfields dataset ~87% accuracy to classify users that will participate in the event.

An increasing number of people share their thoughts and the images of their lives on social media platforms. People are exposed to food in their everyday lives and share on-line what they are eating by means of photos taken to their dishes. The hashtag #foodporn is constantly among the popular hashtags in Twitter and food photos are the second most popular subject in Instagram after selfies. The system that we propose, WorldFoodMap, captures the stream of food photos from social media and, thanks to a CNN food image classifier, identifies the categories of food that people are sharing. By collecting food images from the Twitter stream and associating food category and location to them, WorldFoodMap permits to investigate and interactively visualize the popularity and trends of the shared food all over the world.

Ranking query results effectively by considering user past behaviour and preferences is a primary concern for IR researchers both in academia and industry. In this context, LtR is widely believed to be the most effective solution to design ranking models that account for user-interaction features that have proved to remarkably impact on IR effectiveness. In this paper, we explore the possibility of integrating the user dynamic directly into the LtR algorithms. Specifically, we model with Markov chains the behaviour of users in scanning a ranked result list and we modify Lambdamart, a state-of-the-art LtR algorithm, to exploit a new discount loss function calibrated on the proposed Markovian model of user dynamic. We evaluate the performance of the proposed approach on publicly available LtR datasets, finding that the improvements measured over the standard algorithm are statistically significant.

In this demo paper we propose RankEval, an open-source tool for the analysis and evaluation of Learning-to-Rank (LtR) models based on ensembles of regression trees. Gradient Boosted Regression Trees (GBRT) is a flexible statistical learning technique for classification and regression at the state of the art for training effective LtR solutions. Indeed, the success of GBRT fostered the development of several open-source LtR libraries targeting efficiency of the learning phase and effectiveness of the resulting models. However, these libraries offer only very limited help for the tuning and evaluation of the trained models. In addition, the implementations provided for even the most traditional IR evaluation metrics differ from library to library, thus making the objective evaluation and comparison between trained models a difficult task. RankEval addresses these issues by providing a common ground for LtR libraries that offers useful and interoperable tools for a comprehensive comparison and in-depth analysis of ranking models.

In this paper we propose X-DART, a new Learning to Rank algorithm focusing on the training of robust and compact ranking models. Motivated from the observation that the last trees of MART models impact the prediction of only a few instances of the training set, we borrow from the DART algorithm the dropout strategy consisting in temporarily dropping some of the trees from the ensemble while new weak learners are trained. However, differently from this algorithm we drop permanently these trees on the basis of smart choices driven by accuracy measured on the validation set. Experiments conducted on publicly available datasets shows that X-DART outperforms DART in training models providing the same effectiveness by employing up to 40% less trees.

We propose an analytical framework able to investigate discussions about polarized topics in online social networks from many different angles. The framework supports the analysis of social networks along several dimensions: time, space and sentiment. We show that the proposed analytical framework and the methodology can be used to mine knowledge about the perception of complex social phenomena. We selected the refugee crisis discussions over Twitter as a case study. This difficult and controversial topic is an increasingly important issue for the EU. The raw stream of tweets is enriched with space information (user and mentioned locations), and sentiment (positive vs. negative) w.r.t. refugees. Our study shows differences in positive and negative sentiment in EU countries, in particular in UK, and by matching events, locations and perception, it underlines opinion dynamics and common prejudices regarding the refugees.

The Entity Linking task consists in automatically identifying and linking the entities mentioned in a text to their URIs in a given Knowledge Base, e.g., Wikipedia. Entity Linking has a large im- pact in several text analysis and information retrieval related tasks. This task is very challenging due to natural language ambiguity. However, not all the entities mentioned in a document have the same relevance and utility in understanding the topics being dis- cussed. Thus, the related problem of identifying the most relevant entities present in a document, also known as Salient Entities, is attracting increasing interest. In this paper we propose SEL, a novel supervised two-step algo- rithm comprehensively addressing both entity linking and saliency detection. The first step is based on a classifier aimed at identi- fying a set of candidate entities that are likely to be mentioned in the document, thus maximizing the precision of the method with- out hindering its recall. The second step is still based on machine learning, and aims at choosing from the previous set the entities that actually occur in the document. Indeed, we tested two dif- ferent versions of the second step, one aimed at solving only the entity linking task, and the other that, besides detecting linked en- tities, also scores them according to their saliency. Experiments conducted on two different datasets show that the proposed algo- rithm outperforms state-of-the-art competitors, and is able to detect salient entities with high accuracy.

Several among our daily activities, like traveling to a tourist attraction, are better enjoyed with a group of friends. However, finding the best travel companions is sometimes tricky since we need to form a group of people combining the interest in the proposed destination with the friendship relations among the group members. In this paper we cope with this problem by proposing a new method to recommend the best group of friends with whom to enjoy a specific item, i.e., a travel destination or a venue to visit. Our approach provides a new and original perspective on recommendation: given a user, her social network and a recommended item that is relevant for the user, we want to suggest the best group of friends with whom enjoying the item. This approach differs from traditional group recommendation since it tries to maximize two orthogonal aspects: i) the relevance of the recommended item for every member of the group, and ii), the intra-group social relationships. We introduce the Group Finder framework defining the User-Item Group Formation problem and the possible solutions. We assess our approach in the domain of location recommendation and experiment the proposed solutions using four different publicly available Location Based Social Network (LBSN) datasets. The results achieved confirm the effectiveness and the feasibility of the proposed solutions that outperform strong baselines.

Digital traces of conversations in micro-blogging platforms and OSNs provide information about user opinion with a high degree of resolution. These information sources can be exploited to understand and monitor collective behaviours. In this work, we focus on polarisation classes, i.e., those topics that require the user to side exclusively with one position. The proposed method provides an iterative classification of users and keywords: first, polarised users are identified, then polarised keywords are discovered by monitoring the activities of previously classified users. This method thus allows tracking users and topics over time. We report several experiments conducted on two Twitter datasets during political election time-frames. We measure the user classification accuracy on a golden set of users, and analyse the relevance of the extracted keywords for the ongoing political discussion.

ComeWithMe is an activity oriented carpooling service that enlarges the candidate destinations of a ride request by considering alternative places where the desired activity can be performed. It is based on the observation that individuals often move towards a place to perform an activity while the activity is often not strictly associated with a single place, as one may go for shopping or eating to many different locations. Activity-oriented carpooling hugely increases the number of rides matching a query, thus introducing requirements on system responsiveness and ranking effectiveness that are not common to traditional carpooling services. The demoed system implements the ComeWithMe service in almost its entirety, and includes the back-end and a user-friendly mobile application for smart-phones aimed at achieving users' acceptance and usability.

Learning to Rank (LtR) is the machine learning method of choice for producing high quality document ranking functions from a ground-truth of training examples. In practice, efficiency and effectiveness are intertwined concepts and trading off effectiveness for meeting efficiency constraints typically existing in large-scale systems is one of the most urgent issues. In this paper we propose a new framework, named CLEaVER, for optimizing machine-learned ranking models based on ensembles of regression trees. The goal is to improve efficiency at document scoring time without affecting quality. Since the cost of an ensemble is linear in its size, CLEaVER first removes a subset of the trees in the ensemble, and then fine-tunes the weights of the remaining trees according to any given quality measure. Experiments conducted on two publicly available LtR datasets show that CLEaVER is able to prune up to 80% of the trees and provides an efficiency speed-up up to 2.6x without affecting the effectiveness of the model.

Scoring documents with learning-to-rank (LtR) models based on large ensembles of regression trees is currently deemed one of the best solutions to effectively rank query results to be returned by large scale Information Retrieval systems. This paper investigates the opportunities given by SIMD capabilities of modern CPUs to the end of efficiently evaluating regression trees ensembles. We propose V-QuickScorer (vQS), which exploits SIMD extensions to vectorize the document scoring, i.e., to perform the ensemble traversal by evaluating multiple documents simultaneously. We provide a comprehensive evaluation of vQS against the state of the art on three publicly available datasets. Experiments show that vQS provides speed-ups up to a factor of 3.2x.

Learning-to-Rank models based on additive ensembles of regression trees have been proven to be very effective for scoring query results returned by large-scale Web search engines. Unfortunately, the computational cost of scoring thousands of candidate documents by traversing large ensembles of trees is high. Thus, several works have investigated solutions aimed at improving the efficiency of document scoring by exploiting advanced features of modern CPUs and memory hierarchies. In this article, we present Q uick S corer , a new algorithm that adopts a novel cache-efficient representation of a given tree ensemble, performs an interleaved traversal by means of fast bitwise operations, and supports ensembles of oblivious trees. An extensive and detailed test assessment is conducted on two standard Learning-to-Rank datasets and on a novel very large dataset we made publicly available for conducting significant efficiency tests. The experiments show unprecedented speedups over the best state-of-the-art baselines ranging from 1.9 × to 6.6 × . The analysis of low-level profiling traces shows that Q uick S corer efficiency is due to its cache-aware approach in terms of both data layout and access patterns and to a control flow that entails very low branch mis-prediction rates.

Learning to Rank (LtR) is an effective machine learning methodology for inducing high-quality document ranking functions. Given a query and a candidate set of documents, where query-document pairs are represented by feature vectors, a machine-learned function is used to reorder this set. In this paper we propose a new family of rank-based features, which extend the original feature vector associated with each query-document pair. Indeed, since they are derived as a function of the query-document pair and the full set of candidate documents to score, rank-based features provide additional information to better rank documents and return the most relevant ones. We report a comprehensive evaluation showing that rank-based features allow us to achieve the desired effectiveness with ranking models being up to 3.5 times smaller than models not using them, with a scoring time reduction up to 70%.

Learning-to-Rank models based on additive ensembles of regression trees have proven to be very effective for ranking query results returned by Web search engines, a scenario where quality and efficiency requirements are very demanding. Unfortunately, the computational cost of these ranking models is high. Thus, several works already proposed solutions aiming at improving the efficiency of the scoring process by dealing with features and peculiarities of modern CPUs and memory hierarchies. In this paper, we present QuickScorer, a new algorithm that adopts a novel bitvector representation of the tree-based ranking model, and performs an interleaved traversal of the ensemble by means of simple logical bitwise operations. The performance of the proposed algorithm are unprecedented, due to its cache-aware approach, both in terms of data layout and access patterns, and to a control flow that entails very low branch mis-prediction rates. The experiments on real Learning-to-Rank datasets show that QuickScorer is able to achieve speedups over the best state-of-the-art baseline ranging from 2x to 6.5x.

The Entity Linking (EL) problem consists in automatically linking short fragments of text within a document to entities in a given Knowledge Base like Wikipedia. Due to its impact in several text-understanding related tasks, EL is an hot research topic. The correlated problem of devising the most relevant entities mentioned in the document, a.k.a. salient entities (SE), is also attracting increasing interest. Unfortunately, publicly available evaluation datasets that contain accurate and supervised knowledge about mentioned entities and their relevance ranking are currently very poor both in number and quality. This lack makes very difficult to compare different EL and SE solutions on a fair basis, as well as to devise innovative techniques that relies on these datasets to train machine learning models, in turn used to automatically link and rank entities. In this demo paper we propose a Web-deployed tool that allows to crowdsource the creation of these datasets, by supporting the collaborative human annotation of semi-structured documents. The tool, called Elianto, is actually an open source framework, which provides a user friendly and reactive Web interface to support both EL and SE labelling tasks, through a guided two-step process.

A major mining task for binary matrixes is the extraction of approximate top-k patterns that are able to concisely describe the input data. The top-k pattern discovery problem is commonly stated as an optimization one, where the goal is to minimize a given cost function, see the accuracy of the data description. In this work, we review several greedy algorithms, and discuss PANDA+, an algorithmic framework able to optimize different cost functions generalized into a unifying formulation. We evaluated the goodness of the algorithm by measuring the quality of the extracted patterns. We adapted standard quality measures to assess the capability of the algorithm to discover both the items and transactions of the patterns embedded in the data. The evaluation was conducted on synthetic data, where patterns were artificially embedded, and on real-world text collection, where each document is labeled with a topic. Finally, in order to qualitatively evaluate the usefulness of the discovered patterns, we exploited PANDA+to detect overlapping communities in a bipartite network. The results show that PANDA+is able to discover high-quality patterns in both synthetic and real-world datasets.

In this paper we propose TripBuilder, a new framework for personalized touristic tour planning. We mine from Flickr the information about the actual itineraries followed by a multitude of different tourists, and we match these itineraries on the touristic Point of Interests available from Wikipedia. The task of planning personalized touristic tours is then modeled as an instance of the Generalized Maximum Coverage problem. Wisdom-of-the-crowds information allows us to derive touristic plans that maximize a measure of interest for the tourist given her preferences and visiting time-budget. Experimental results on three different touristic cities show that our approach is effective and outperforms strong baselines.

A search engine infrastructure must be able to provide the same quality of service to all queries received during a day. During normal operating conditions, the demand for resources is considerably lower than under peak conditions, yet an oversized infrastructure would result in an unnecessary waste of computing power. A possible solution adopted in this situation might consist of defining a maximum threshold processing time for each query, and dropping queries for which this threshold elapses, leading to disappointed users. In this paper, we propose and evaluate a different approach, where, given a set of different query processing strategies with differing efficiency, each query is considered by a framework that sets a maximum query processing time and selects which processing strategy is the best for that query, such that the processing time for all queries is kept below the threshold. The processing time estimates used by the scheduler are learned from past queries. We experimentally validate our approach on 10,000 queries from a standard TREC dataset with over 50 million documents, and we compare it with several baselines. These experiments encompass testing the system under different query loads and different maximum tolerated query response times. Our results show that, at the cost of a marginal loss in terms of response quality, our search system is able to answer 90% of queries within half a second during times of high query volume.

Entity Linking is the task of detecting, in text documents, relevant mentions to entities of a given knowledge base. To this end, entity-linking algorithms use several signals and features extracted from the input text or from the knowledge base. The most important of such features is entity relatedness. Indeed, we argue that these algorithms benefit from maximizing the relatedness among the relevant entities selected for annotation, since this minimizes errors in disambiguating entity-linking.

Although Web search engines still answer user queries with lists of ten blue links to webpages, people are increasingly issuing queries to accomplish their daily tasks (e.g., finding a recipe , booking a flight , reading online news , etc.). In this work, we propose a two-step methodology for discovering tasks that users try to perform through search engines. First, we identify user tasks from individual user sessions stored in search engine query logs. In our vision, a user task is a set of possibly noncontiguous queries (within a user search session), which refer to the same need. Second, we discover collective tasks by aggregating similar user tasks, possibly performed by distinct users. To discover user tasks, we propose query similarity functions based on unsupervised and supervised learning approaches. We present a set of query clustering methods that exploit these functions in order to detect user tasks. All the proposed solutions were evaluated on a manually-built ground truth, and two of them performed better than state-of-the-art approaches. To detect collective tasks, we propose four methods that cluster previously discovered user tasks, which in turn are represented by the bag-of-words extracted from their composing queries. These solutions were also evaluated on another manually-built ground truth.

Recommender systems have become ubiquitous in content-based web applications, from news to shopping sites. Nonetheless, an aspect that has been largely overlooked so far in the recommender system literature is that of automatically building explanations for a particular recommendation. This paper focuses on the news domain, and proposes to enhance effectiveness of news recommender systems by adding, to each recommendation, an explanatory statement to help the user to better understand if, and why, the item can be her interest. We consider the news recommender system as a black-box, and generate different types of explanations employing pieces of information associated with the news. In particular, we engineer text-based, entity-based, and usage-based explanations, and make use of a Markov Logic Networks to rank the explanations on the basis of their effectiveness. The assessment of the model is conducted via a user study on a dataset of news read consecutively by actual users. Experiments show that news recommender systems can greatly benefit from our explanation module as it allows users to discriminate between interesting and not interesting news in the majority of the cases.

We present a recommendation method based on the well-known concept of center-piece subgraph, that allows for the time/space efficient generation of suggestions also for rare, i.e., long-tail queries. Our method is scalable with respect to both the size of datasets from which the model is computed and the heavy workloads that current web search engines have to deal with. Basically, we relate terms contained into queries with highly correlated queries in a query-flow graph. This enables a novel recommendation generation method able to produce recommendations for approximately 99% of the workload of a real-world search engine. The method is based on a graph having term nodes, query nodes, and two kinds of connections: term-query and query-query. The first connects a term to the queries in which it is contained, the second connects two query nodes if the likelihood that a user submits the second query after having issued the first one is sufficiently high. On such large graph we need to compute the center-piece subgraph induced by terms contained into queries. In order to reduce the cost of the above computation, we introduce a novel and efficient method based on an inverted index representation of the model. We experiment our solution on two real-world query logs and we show that its effectiveness is comparable (and in some case better) than state-of-the-art methods for head-queries. More importantly, the quality of the recommendations generated remains very high also for long-tail queries, where other methods fail even to produce any suggestion. Finally, we extensively investigate scalability and efficiency issues and we show the viability of our method in real world search engines.

Web Search Engines ’ result pages contain references to the top-k documents relevant for the query submitted by a user. Each document is represented by a title, a snippet and a URL. Snippets, i.e. short sentences showing the portions of the document being relevant to the query, help users to select the most interesting results. The snippet generation process is very expensive, since it may require to access a number of documents for each issued query. We assert that caching, a popular technique used to enhance performance at various levels of any computing sys-tems, can be very effective in this context. We design and experiment several cache organizations, and we introduce the concept of supersnippet, that is the set of sentences in a document that are more likely to answer future queries. We show that supersnippets can be built by exploiting query logs, and that in our experiments a supersnippet cache an-swers up to 62 % of the requests, remarkably outperforming other caching approaches.

The research challenge addressed in this paper is to devise effective techniques for identifying task-based sessions, i.e. sets of possibly non contiguous queries issued by the user of a Web Search Engine for carrying out a given task. In order to evaluate and compare different approaches, we built, by means of a manual labeling process, a ground-truth where the queries of a given query log have been grouped in tasks. Our analysis of this ground-truth shows that users tend to perform more than one task at the same time, since about 75% of the submitted queries involve a multi-tasking activity. We formally define the Task-based Session Discovery Problem (TSDP) as the problem of best approximating the manually annotated tasks, and we propose several variants of well known clustering algorithms, as well as a novel efficient heuristic algorithm, specifically tuned for solving the TSDP. These algorithms also exploit the collaborative knowledge collected by Wiktionary and Wikipedia for detecting query pairs that are not similar from a lexical content point of view, but actually semantically related. The proposed algorithms have been evaluated on the above ground-truth, and are shown to perform better than state-of-the-art approaches, because they effectively take into account the multi-tasking behavior of users.

In this paper we analyze the efficiency of various search results diversification methods. While efficacy of diversification approaches has been deeply investigated in the past, response time and scalability issues have been rarely addressed. A unified framework for studying performance and feasibility of result diversification solutions is thus proposed. First we define a new methodology for detecting when, and how, query results need to be diversified. To this purpose, we rely on the concept of "query refinement" to estimate the probability of a query to be ambiguous . Then, relying on this novel ambiguity detection method, we deploy and compare on a standard test set, three different diversification methods: IASelect, xQuAD, and OptSelect. While the first two are recent state-of-the-art proposals, the latter is an original algorithm introduced in this paper. We evaluate both the efficiency and the effectiveness of our approach against its competitors by using the standard TREC Web diversification track testbed. Results shown that OptSelect is able to run two orders of magnitude faster than the two other state-of-the-art approaches and to obtain comparable figures in diversification effectiveness.

Given a set of records, a threshold value t and a similarity function, we investigate the problem of finding all pairs of records such that similarity between each pair is above t. We propose several optimizations on the existing approaches to solve the problem. Our algorithm outperforms the state-of-the-art algorithms in the case with large and high-dimensional datasets. The speedup we achieved varied from 30% to 4-x depending on the similarity threshold and the dataset properties.

The discovery of patterns in binary dataset has many applications, e.g. in electronic commerce, TCP/IP networking, Web usage logging, etc. Still, this is a very challenging task in many respects: overlapping vs. non overlapping patterns, presence of noise, extraction of the most important patterns only. In this paper we formalize the problem of discovering the Top-K patterns from binary datasets in presence of noise, as the minimization of a novel cost function. According to the Minimum Description Length principle, the proposed cost function favors succinct pattern sets that may approximately describe the input data. We propose a greedy algorithm for the discovery of Patterns in Noisy Datasets, named PaNDa, and show that it outperforms related techniques on both synthetic and real-world data.

Static index pruning techniques aim at removing from the posting lists of an inverted file the references to documents which are likely to be not relevant for answering user queries. The reduction in the size of the index results in a better exploitation of memory hierarchies and faster query processing. On the other hand, pruning may affect the precision of the information retrieval system, since pruned entries are unavailable at query processing time. Static pruning techniques proposed so far exploit query-independent measures to evaluate the importance of a document within a posting list. This paper proposes a general framework that aims at enhancing the precision of any static pruning methods by exploiting usage information extracted from query logs. Experiments conducted on the TREC WT10g Web collection and a large Altavista query log show that integrating usage knowledge into the pruning process is profitable, and increases remarkably performance figures obtained with the state-of-the art Carmel's static pruning method.

This article introduces an architecture for a document-partitioned search engine, based on a novel approach combining collection selection and load balancing, called load-driven routing . By exploiting the query-vector document model, and the incremental caching technique, our architecture can compute very high quality results for any query, with only a fraction of the computational load used in a typical document-partitioned architecture. By trading off a small fraction of the results, our technique allows us to strongly reduce the computing pressure to a search engine back-end; we are able to retrieve more than 2/3 of the top-5 results for a given query with only 10% the computing load needed by a configuration where the query is processed by each index partition. Alternatively, we can slightly increase the load up to 25% to improve precision and get more than 80% of the top-5 results. In fact, the flexibility of our system allows a wide range of different configurations, so as to easily respond to different needs in result quality or restrictions in computing power. More important, the system configuration can be adjusted dynamically in order to fit unexpected query peaks or unpredictable failures. This article wraps up some recent works by the authors, showing the results obtained by tests conducted on 6 million documents, 2,800,000 queries and real query cost timing as measured on an actual index.

World Wide Web content continuously grows in size and importance. Furthermore, users ask Web search engines to satisfy increasingly disparate information needs. New techniques and tools are constantly developed aimed at assisting users in the interaction with the Web search engine. Query recommender systems suggesting interesting queries to users are an example of such tools. Most query recommendation techniques are based on the knowledge of the behaviors of past users of the search engine recorded in query logs.

In order to become an effective complement to traditional Web-scale text-based image retrieval solutions, content-based image retrieval must address scalability and efficiency issues. In this paper we investigate the possibility of caching the answers to content-based image retrieval queries in metric space, with the aim of reducing the average cost of query processing, and boosting the overall system throughput. Our proposal exploits the similarity between the query object and the cache content, and allows the cache to return approximate answers with acceptable quality guarantee even if the query processed has never been encountered in the past. Moreover, since popular images that are likely to be used as query have several near-duplicate versions, we show that our caching algorithm is robust, and does not suffer of cache pollution problems due to near-duplicate query objects. We report on very promising results obtained with a collection of one million high-quality digital photos. We show that it is worth pursuing caching strategies also in similarity search systems, since the proposed caching techniques can have a significant impact on performance, like caching on text queries has been proven effective for traditional Web search engines.

The recommendation of queries, known as query suggestion, is a common practice on major Web Search Engines. It aims to help users to find the information they are looking for, and is usually based on the knowledge learned from past interactions with the search engine. In this paper we propose a new model for query suggestion, the Search Shortcut Problem, that consists in recommending "successful" queries that allowed other users to satisfy, in the past, similar information needs. This new model has several advantages with respect to traditional query suggestion approaches. First, it allows a straightforward evaluation of algorithms from available query log data. Moreover, it simplifies the application of several recommendation techniques from other domains. Particularly, in this work we applied Collaborative Filtering to this problem, and evaluated the interesting results achieved on large query logs from AOL and Microsoft. Different techniques for analyzing and extracting information from query logs, as well as new metrics and techniques for measuring the effectiveness of recommendations are proposed and evaluated. The results obtained clearly show the importance of several of our contributions, and open an interesting field for future research.

Giving suggestions to users of Web-based services is a common practice aimed at enhancing their navigation experience. Major Web Search Engines usually provide "Suggestions" under the form of queries that are, to some extent, related to the current query typed by the user, and the knowledge learned from the past usage of the system. In this work we introduce "Search Shortcuts" as "Successful" queries allowed, in the past, users to satisfy their information needs. Differently from conventional suggestion techniques, our search shortcuts allows to evaluate effectiveness by exploiting a simple train-and-test approach. We have applied several Collaborative Filtering algorithms to this problem, evaluating them on a real query log data. We generate the shortcuts from all user sessions belonging to the testing set, and measure the quality of the shortcuts suggested by considering the similarity between them and the navigational user behavior.

Inspired by emerging multi-core computer architectures, in this paper we present MT CL O S E D, a multi-threaded algorithm for frequent closed itemset mining (FCIM). To the best of our knowledge, this is the first FCIM parallel algorithm proposed so far. We studied how different duplicate checking techniques, typical of FCIM algorithms, may affect this parallelization. We showed that only one of them allows to decompose the global FCIM problem into independent tasks that can be executed in any order, and thus in parallel. Finally we show how MT CL O S E D efficiently harness modern CPUs. We designed and tested several parallelization paradigms by investigating static/dynamic decomposition and scheduling of tasks, thus showing its scalability w.r.t. to the number of CPUs. We analyzed the cache friendliness of the algorithm. Finally, we provided additional speed-up by introducing SIMD extensions.

ConQueSt is a constraint-based querying system devised with the aim of supporting the intrinsically exploratory nature of pattern discovery. It provides users with an expressive constraint-based query language which allows the discovery process to be effectively driven toward potentially interesting patterns. Constraints are also exploited to reduce the cost of pattern mining. The system is built around an efficient constraint-based mining engine which entails several data and search space reduction techniques, and allows new user-defined constraints to be easily added.

Extracting frequent itemsets is an important task in many data mining applications. When data are very large, it becomes mandatory to perform the mining task by using an external memory algorithm, but only a few of these algorithms have been proposed so far. Since also the result set of all the frequent itemsets is likely to be undesirably large, condensed representations, such as closed itemsets, have recently gained a lot of attention. In this paper we discuss the limitations of the partitioning techniques adopted by external memory algorithms for extracting all the frequent itemsets, when applied to closed itemsets mining. The main issue is that the closedness of an itemset cannot be evaluated only using the local knowledge available in a single partition of the input dataset. A further step is thus needed to correctly merge the partial results. We introduce the first algorithm for mining closed itemsets out of core. The algorithm exploits a divide-et-impera approach, where the input dataset is split into smaller partitions, such that not only they can be loaded, but also they can be mined entirely into the main memory. Moreover, we devised a simple technique based on a new theoretical result that allows us to reduce the problem of merging partial solutions to an external memory sorting problem.

This paper presents a new scalable algorithm for discovering closed frequent itemsets, a lossless and condensed representation of all the frequent itemsets that can be mined from a transactional database. Our algorithm exploits a divide-and-conquer approach and a bitwise vertical representation of the database and adopts a particular visit and partitioning strategy of the search space based on an original theoretical framework, which formalizes the problem of closed itemsets mining in detail. The algorithm adopts several optimizations aimed to save both space and time in computing itemset closures and their supports. In particular, since one of the main problems in this type of algorithms is the multiple generation of the same closed itemset, we propose a new effective and memory-efficient pruning technique, which, unlike other previous proposals, does not require the whole set of closed patterns mined so far to be kept in the main memory. This technique also permits each visited partition of the search space to be mined independently in any order and, thus, also in parallel. The tests conducted on many publicly available data sets show that our algorithm is scalable and outperforms other state-of-the-art algorithms like CLOSET+ and FP-CLOSE, in some cases by more than one order of magnitude. More importantly, the performance improvements become more and more significant as the support threshold is decreased.

This article discusses efficiency and effectiveness issues in caching the results of queries submitted to a Web search engine (WSE). We propose SDC (Static Dynamic Cache), a new caching strategy aimed to efficiently exploit the temporal and spatial locality present in the stream of processed queries. SDC extracts from historical usage data the results of the most frequently submitted queries and stores them in astatic,read-onlyportion of the cache. The remaining entries of the cache are dynamically managed according to a given replacement policy and are used for those queries that cannot be satisfied by the static portion. Moreover, we improve the hit ratio of SDC by using an adaptive prefetching strategy, which anticipates future requests by introducing a limited overhead over the back-end WSE. We experimentally demonstrate the superiority of SDC over purely static and dynamic policies by measuring the hit ratio achieved on three large query logs by varying the cache parameters and the replacement policy used for managing the dynamic part of the cache. Finally, we deploy and measure the throughput achieved by a concurrent version of our caching system. Our tests show how the SDC cache can be efficiently exploited by many threads that concurrently serve the queries of different users.

Web Search Engines provide a large-scale text document retrieval service by processing huge Inverted File indexes. Inverted File indexes allow fast query resolution and good memory utilization since their d-gaps representation can be effectively and efficiently compressed by using variable length encoding methods. This paper proposes and evaluates some algorithms aimed to find an assignment of the document identifiers which minimizes the average values of d-gaps, thus enhancing the effectiveness of traditional compression methods. We ran several tests over the Google contest collection in order to validate the techniques proposed. The experiments demonstrated the scalability and effectiveness of our algorithms. Using the proposed algorithms, we were able to sensibly improve (up to 20.81%) the compression ratios of several encoding schemes.

The performance of an algorithm that mines frequent sets from transactional databases may severely depend on the specific features of the data being analyzed. Moreover, some architectural characteristics of the computational platform used - e.g. the available main memory - can dramatically change its runtime behavior. In this paper we present DCI (Direct Count & Intersect), an efficient algorithm for discovering frequent sets from large databases. Due to the multiple heuristics strategies adopted, DCI can adapt its behavior not only to the features of the specific computing platform, but also to the features of the dataset being mined, so that it results very effective in mining both short and long patterns from sparse and dense datasets. Finally we also discuss the parallelization strategies adopted in the design of ParDCI, a distributed and multi-threaded implementation of DCI.