LLM-powered search agents are increasingly being used for multi-step information seeking tasks, yet the IR community lacks empirical understanding of how agentic search sessions unfold and how retrieved evidence is reflected in later queries. This paper presents a large-scale log analysis of agentic search based on 14.44M search requests (3.97M sessions) collected from DeepResearchGym, i.e., an open-source search API accessed by external agentic clients. We sessionize the logs, assign session-level intents and step-wise query-reformulation labels using LLM-based annotation, and propose Context-driven Term Adoption Rate (CTAR) to quantify whether newly introduced query terms are lexically traceable to previously retrieved evidence. Our analyses reveal distinctive behavioral patterns. First, over 90\% of multi-turn sessions contain at most ten steps, and 89\% of inter-step intervals fall under one minute. Second, behavior varies by intent. Fact-seeking sessions exhibit high repetition that increases over time, while sessions requiring reasoning sustain broader exploration. Third, query reformulations are often traceable to retrieved evidence across steps. On average, 54\% of newly introduced query terms appear in the accumulated evidence context, with additional traceability to earlier steps beyond the most recent retrieval. These findings provide candidate signals for repetition-aware stopping, intent-adaptive retrieval budgeting, and explicit cross-step context tracking. We released the anonymized logs, making them available at a public HuggingFace~\chref{https://huggingface.co/datasets/cx-cmu/deepresearchgym-agentic-search-logs}{repository}.

Recent advancements in large language models (LLMs) have driven interest in billion-scale retrieval models with strong generalization across retrieval tasks and languages. Additionally, progress in large vision-language models has created new opportunities for multimodal retrieval. In response, we have updated the Tevatron toolkit, introducing a unified pipeline that enables researchers to explore retriever models at different scales, across multiple languages, and with various modalities. This demo paper highlights the toolkit's key features, bridging academia and industry by supporting efficient training, inference, and evaluation of neural retrievers. We showcase a unified dense retriever achieving strong multilingual and multimodal effectiveness, and conduct a cross-modality zero-shot study to demonstrate its research potential. Alongside, we release OmniEmbed, to the best of our knowledge, the first embedding model that unifies text, image document, video, and audio retrieval, serving as a baseline for future research.

Tip-of-the-Tongue (ToT) retrieval is challenging for search engines because the queries are usually natural-language, verbose, and contain uncertain and inaccurate information. This paper studies the generalization capabilities of existing retrieval methods with ToT queries in multiple domains. We curate a multi-domain dataset and evaluate the effectiveness of recall-oriented first-stage retrieval methods across the different domains, considering in-domain, out-of-domain, and multi-domain training settings. We further explore the use of a Large Language Model (LLM), i.e. GPT-4, for zero-shot re-ranking in various ToT domains, relying solely on the item titles. Results show that multi-domain training enhances recall, and that LLMs are strong zero-shot re-rankers, especially for popular items, outperforming direct GPT-4 prompting without first-stage retrieval. Datasets and code can be found on GitHub https://github.com/LuisPB7/TipTongue

Recent rapid advances in deep pre-trained language models and the introduction of large datasets have powered research in embedding-based neural retrieval. While many excellent research papers have emerged, most of them come with their own implementations, which are typically optimized for some particular research goals instead of efficiency or code organization. In this paper, we introduce Tevatron, a neural retrieval toolkit that is optimized for efficiency, flexibility, and code simplicity. Tevatron enables model training and evaluation for a variety of ranking components such as dense retrievers, sparse retrievers, and rerankers. It also provides a standardized pipeline that includes text processing, model training, corpus/query encoding, and search. In addition, Tevatron incorporates well-studied methods for improving retriever effectiveness such as hard negative mining and knowledge distillation. We provide an overview of Tevatron in this paper, demonstrating its effectiveness and efficiency on multiple IR and QA datasets. We highlight Tevatron's flexible design, which enables easy generalization across datasets, model architectures, and accelerator platforms (GPUs and TPUs). Overall, we believe that Tevatron can serve as a solid software foundation for research on neural retrieval systems, including their design, modeling, and optimization.

Long document re-ranking has been a challenging problem for neural re-rankers based on deep language models like BERT. Early work breaks the documents into short passage-like chunks. These chunks are independently mapped to scalar scores or latent vectors, which are then pooled into a final relevance score. These encode-and-pool methods however inevitably introduce an information bottleneck: the low dimension representations. In this paper, we propose instead to model full query-to-document interaction, leveraging the attention operation and modular Transformer re-ranker framework. First, document chunks are encoded independently with an encoder module. An interaction module then encodes the query and performs joint attention from the query to all document chunk representations. We demonstrate that the model can use this new degree of freedom to aggregate important information from the entire document. Our experiments show that this design produces effective re-ranking on two classical IR collections Robust04 and ClueWeb09, and a large-scale supervised collection MS-MARCO document ranking.

ClueWeb22, the newest iteration of the ClueWeb line of datasets, is the result of more than a year of collaboration between industry and academia. Its design is influenced by the research needs of the academic community and the real-world needs of large-scale industry systems. Compared with earlier ClueWeb datasets, the ClueWeb22 corpus is larger, more varied, and has higher-quality documents. Its core is raw HTML, but it includes clean text versions of documents to lower the barrier to entry. Several aspects of ClueWeb22 are available to the research community for the first time at this scale, for example, visual representations of rendered web pages, parsed structured information from the HTML document, and the alignment of document distributions (domains, languages, and topics) to commercial web search.

Dense retrieval systems conduct first-stage retrieval using embedded representations and simple similarity metrics to match a query to documents. Its effectiveness depends on encoded embeddings to capture the semantics of queries and documents, a challenging task due to the shortness and ambiguity of search queries. This paper proposes ANCE-PRF, a new query encoder that uses pseudo relevance feedback (PRF) to improve query representations for dense retrieval. ANCE-PRF uses a BERT encoder that consumes the query and the top retrieved documents from a dense retrieval model, ANCE, and it learns to produce better query embeddings directly from relevance labels. It also keeps the document index unchanged to reduce overhead. ANCE-PRF significantly outperforms ANCE and other recent dense retrieval systems on several datasets. Analysis shows that the PRF encoder effectively captures the relevant and complementary information from PRF documents, while ignoring the noise with its learned attention mechanism.

Category systems are central components of knowledge bases, as they provide a hierarchical grouping of semantically related concepts and entities. They are a unique and valuable resource that is utilized in a broad range of information access tasks. To aid knowledge editors in the manual process of expanding a category system, this paper presents a method of generating categories for sets of entities. First, we employ neural abstractive summarization models to generate candidate categories. Next, the location within the hierarchy is identified for each candidate. Finally, structure-, content-, and hierarchy-based features are used to rank candidates to identify by the most promising ones (measured in terms of specificity, hierarchy, and importance). We develop a test collection based on Wikipedia categories and demonstrate the effectiveness of the proposed approach.

Given a natural language query, teaching machines to ask clarifying questions is of immense utility in practical natural language processing systems. Such interactions could help in filling information gaps for better machine comprehension of the query. For the task of ranking clarification questions, we hypothesize that determining whether a clarification question pertains to a missing entry in a given post (on QA forums such as StackExchange) could be considered as a special case of Natural Language Inference (NLI), where both the post and the most relevant clarification question point to a shared latent piece of information or context. We validate this hypothesis by incorporating representations from a Siamese BERT model fine-tuned on NLI and Multi-NLI datasets into our models and demonstrate that our best performing model obtains a relative performance improvement of 40 percent and 60 percent respectively (on the key metric of [email protected]), over the state-of-the-art baseline(s) on the two evaluation sets of the StackExchange dataset, thereby, significantly surpassing the state-of-the-art.

CAsT-19 is a new dataset that supports research on conversational information seeking. The corpus is 38,426,252 passages from the TREC Complex Answer Retrieval (CAR) and Microsoft MAchine Reading COmprehension (MARCO) datasets. Eighty information seeking dialogues (30 train, 50 test) are an average of 9 to 10 questions long. A dialogue may explore a topic broadly or drill down into subtopics. Questions contain ellipsis, implied context, mild topic shifts, and other characteristics of human conversation that may prevent them from being understood in isolation. Relevance assessments are provided for 30 training topics and 20 test topics.

Tabular data provide answers to a significant portion of search queries. However, reciting an entire result table is impractical in conversational search systems. We propose to generate natural language summaries as answers to describe the complex information contained in a table. Through crowdsourcing experiments, we build a new conversation-oriented, open-domain table summarization dataset. It includes annotated table summaries, which not only answer questions but also help people explore other information in the table. We utilize this dataset to develop automatic table summarization systems as SOTA baselines. Based on the experimental results, we identify challenges and point out future research directions that this resource will support.

Term frequency is a common method for identifying the importance of a term in a document. But term frequency ignores how a term interacts with its text context, which is key to estimating document-specific term weights. This paper proposes a Deep Contextualized Term Weighting framework (DeepCT) that maps the contextualized term representations from BERT to into context-aware term weights for passage retrieval. The new, deep term weights can be stored in an ordinary inverted index for efficient retrieval. Experiments on two datasets demonstrate that DeepCT greatly improves the accuracy of first-stage passage retrieval algorithms.

Language model pre-training has spurred a great deal of attention for tasks involving natural language understanding, and has been successfully applied to many downstream tasks with impressive results. Within information retrieval, many of these solutions are too costly to stand on their own, requiring multi-stage ranking architectures. Recent work has begun to consider how to "backport" salient aspects of these computationally expensive models to previous stages of the retrieval pipeline. One such instance is DeepCT, which uses BERT to re-weight term importance in a given context at the passage level. This process, which is computed offline, results in an augmented inverted index with re-weighted term frequency values. In this work, we conduct an investigation of query processing efficiency over DeepCT indexes. Using a number of candidate generation algorithms, we reveal how term re-weighting can impact query processing latency, and explore how DeepCT can be used as a static index pruning technique to accelerate query processing without harming search effectiveness.

Bag-of-words document representations play a fundamental role in modern search engines, but their power is limited by the shallow frequency-based term weighting scheme. This paper proposes HDCT, a context-aware document term weighting framework for document indexing and retrieval. It first estimates the semantic importance of a term in the context of each passage. These fine-grained term weights are then aggregated into a document-level bag-of-words representation, which can be stored into a standard inverted index for efficient retrieval. This paper also proposes two approaches that enable training HDCT without relevance labels. Experiments show that an index using HDCT weights significantly improved the retrieval accuracy compared to typical term-frequency and state-of-the-art embedding-based indexes.

Neural networks provide new possibilities to automatically learn complex language patterns and query-document relations. Neural IR models have achieved promising results in learning query-document relevance patterns, but few explorations have been done on understanding the text content of a query or a document. This paper studies leveraging a recently-proposed contextual neural language model, BERT, to provide deeper text understanding for IR. Experimental results demonstrate that the contextual text representations from BERT are more effective than traditional word embeddings. Compared to bag-of-words retrieval models, the contextual language model can better leverage language structures, bringing large improvements on queries written in natural languages. Combining the text understanding ability with search knowledge leads to an enhanced pre-trained BERT model that can benefit related search tasks where training data are limited.

Newsworthy events are broadcast through multiple mediums and prompt the crowds to produce comments on social media. In this paper, we propose to leverage on this behavioral dynamics to estimate the most relevant time periods for an event (i.e., query). Recent advances have shown how to improve the estimation of the temporal relevance of such topics. In this approach, we build on two major novelties. First, we mine temporal evidences from hundreds of external sources into topic-based external collections to improve the robustness of the detection of relevant time periods. Second, we propose a formal retrieval model that generalizes the use of the temporal dimension across different aspects of the retrieval process. In particular, we show that temporal evidence of external collections can be used to (i) infer a topic's temporal relevance, (ii) select the query expansion terms, and (iii) re-rank the final results for improved precision. Experiments with TREC Microblog collections show that the proposed time-aware retrieval model makes an effective and extensive use of the temporal dimension to improve search results over the most recent temporal models. Interestingly, we observe a strong correlation between precision and the temporal distribution of retrieved and relevant documents.

Finding diagrams that contain a specific part or a similar part is important in many engineering tasks. In this search task, the query part is expected to match only a small region in a complex image. This paper investigates several local matching networks that explicitly model local region-to-region similarities. Deep convolutional neural networks extract local features and model local matching patterns. Spatial convolution is employed to cross-match local regions at different scale levels, addressing cases where the target part appears at a different scale, position, and/or angle. A gating network automatically learns region importance, removing noise from sparse areas and visual metadata in engineering diagrams.

Selective search architectures use resource selection algorithms such as Rank-S or Taily to rank index shards and determine how many to search for a given query. Most prior research evaluated solutions by their ability to improve efficiency without significantly reducing early-precision metrics such as [email protected] and [email protected] This paper recasts selective search as an early stage of a multi-stage retrieval architecture, which makes recall-oriented metrics more appropriate. A new algorithm is presented that predicts the number of shards that must be searched for a given query in order to meet recall-oriented goals. Decoupling shard ranking from deciding how many shards to search clarifies efficiency vs. effectiveness trade-offs, and enables them to be optimized independently. Experiments on two corpora demonstrate the value of this approach.

This paper studies the consistency of the kernel-based neural ranking model K-NRM, a recent state-of-the-art neural IR model, which is important for reproducible research and deployment in the industry. We find that K-NRM has low variance on relevance-based metrics across experimental trials. In spite of this low variance in overall performance, different trials produce different document rankings for individual queries. The main source of variance in our experiments was found to be different latent matching patterns captured by K-NRM. In the IR-customized word embeddings learned by K-NRM, the query-document word pairs follow two different matching patterns that are equally effective, but align word pairs differently in the embedding space. The different latent matching patterns enable a simple yet effective approach to construct ensemble rankers, which improve K-NRM's effectiveness and generalization abilities.

This paper presents a Kernel Entity Salience Model (KESM) that improves text understanding and retrieval by better estimating entity salience (importance) in documents. KESM represents entities by knowledge enriched distributed representations, models the interactions between entities and words by kernels, and combines the kernel scores to estimate entity salience. The whole model is learned end-to-end using entity salience labels. The salience model also improves ad hoc search accuracy, providing effective ranking features by modeling the salience of query entities in candidate documents. Our experiments on two entity salience corpora and two TREC ad hoc search datasets demonstrate the effectiveness of KESM over frequency-based and feature-based methods. We also provide examples showing how KESM conveys its text understanding ability learned from entity salience to search.

This paper presents \textttConv-KNRM, a Convolutional Kernel-based Neural Ranking Model that models n-gram soft matches for ad-hoc search. Instead of exact matching query and document n-grams, \textttConv-KNRM uses Convolutional Neural Networks to represent n-grams of various lengths and soft matches them in a unified embedding space. The n-gram soft matches are then utilized by the kernel pooling and learning-to-rank layers to generate the final ranking score. \textttConv-KNRM can be learned end-to-end and fully optimized from user feedback. The learned model»s generalizability is investigated by testing how well it performs in a related domain with small amounts of training data. Experiments on English search logs, Chinese search logs, and TREC Web track tasks demonstrated consistent advantages of \textttConv-KNRM over prior neural IR methods and feature-based methods.

Entity-based ranking systems often employ entity linking systems to align entities to query and documents. Previously, entity linking systems were not designed specifically for search engines and were mostly used as a preprocessing step. This work presents JointSem, a joint semantic ranking system that combines query entity linking and entity-based document ranking. In JointSem, the spotting and linking signals are used to describe the importance of candidate entities in the query, and the linked entities are utilized to provide additional ranking features for the documents. The linking signals and the ranking signals are combined by a joint learning-to-rank model, and the whole system is fully optimized towards end-to-end ranking performance. Experiments on TREC Web Track datasets demonstrate the effectiveness of joint learning of entity linking and entity-based ranking.

We present a learning-to-rank approach for resource selection. We develop features for resource ranking and present a training approach that does not require human judgments. Our method is well-suited to environments with a large number of resources such as selective search, is an improvement over the state-of-the-art in resource selection for selective search, and is statistically equivalent to exhaustive search even for recall-oriented metrics such as [email protected], an area in which selective search was lacking.

The DBpedia-entity collection has been used as a standard test collection for entity search in recent years. We develop and release a new version of this test collection, DBpedia-Entity v2, which uses a more recent DBpedia dump and a unified candidate result pool from the same set of retrieval models. Relevance judgments are also collected in a uniform way, using the same group of crowdsourcing workers, following the same assessment guidelines. The result is an up-to-date and consistent test collection.To facilitate further research, we also provide details about the pre-processing and indexing steps, and include baseline results from both classical and recently developed entity search methods.

This paper presents a word-entity duet framework for utilizing knowledge bases in ad-hoc retrieval. In this work, the query and documents are modeled by word-based representations and entity-based representations. Ranking features are generated by the interactions between the two representations, incorporating information from the word space, the entity space, and the cross-space connections through the knowledge graph. To handle the uncertainties from the automatically constructed entity representations, an attention-based ranking model AttR-Duet is developed. With back-propagation from ranking labels, the model learns simultaneously how to demote noisy entities and how to rank documents with the word-entity duet. Evaluation results on TREC Web Track ad-hoc task demonstrate that all of the four-way interactions in the duet are useful, the attention mechanism successfully steers the model away from noisy entities, and together they significantly outperform both word-based and entity-based learning to rank systems.

This paper proposes K-NRM, a kernel based neural model for document ranking. Given a query and a set of documents, K-NRM uses a translation matrix that models word-level similarities via word embeddings, a new kernel-pooling technique that uses kernels to extract multi-level soft match features, and a learning-to-rank layer that combines those features into the final ranking score. The whole model is trained end-to-end. The ranking layer learns desired feature patterns from the pairwise ranking loss. The kernels transfer the feature patterns into soft-match targets at each similarity level and enforce them on the translation matrix. The word embeddings are tuned accordingly so that they can produce the desired soft matches. Experiments on a commercial search engine's query log demonstrate the improvements of K-NRM over prior feature-based and neural-based states-of-the-art, and explain the source of K-NRM's advantage: Its kernel-guided embedding encodes a similarity metric tailored for matching query words to document words, and provides effective multi-level soft matches.

Many URLs on the Internet point to identical contents, which increase the burden of web crawlers. Techniques that detect such URLs (known as URL de-duping) can greatly save resources such as bandwidth and storage for crawlers. Traditional de-duping methods are usually limited to heavily engineered rule matching strategies.In this work, we propose a novel URL de-duping framework based on sequence-to-sequence (Seq2Seq) neural networks. A single concise translation model can take the place of thousands of explicit rules. Experiments indicate that a vanilla Seq2Seq architecture yields robust and accurate results in detecting duplicate URLs. Furthermore, we demonstrate the efficiency of this framework in the real large-scale web environment.

This paper introduces Explicit Semantic Ranking (ESR), a new ranking technique that leverages knowledge graph embedding. Analysis of the query log from our academic search engine, SemanticScholar.org, reveals that a major error source is its inability to understand the meaning of research concepts in queries. To addresses this challenge, ESR represents queries and documents in the entity space and ranks them based on their semantic connections from their knowledge graph embedding. Experiments demonstrate ESR's ability in improving Semantic Scholar's online production system, especially on hard queries where word-based ranking fails.

Selective search is a cluster-based distributed retrieval architecture that reduces computational costs by partitioning a corpus into topical shards, and selectively searching them. Prior research formed topical shards by clustering the corpus based on the documents' contents. This content-based partitioning strategy reveals common topics in a corpus. However, the topic distribution produced by clustering may not match the distribution of topics in search traffic, which may reduce the effectiveness of selective search.

This work investigates the effectiveness of learning to rank methods for entity search. Entities are represented by multi-field documents constructed from their RDF triples, and field-based text similarity features are extracted for query-entity pairs. State-of-the-art learning to rank methods learn models for ad-hoc entity search. Our experiments on an entity search test collection based on DBpedia confirm that learning to rank methods are as powerful for ranking entities as for ranking documents, and establish a new state-of-the-art for accuracy on this benchmark dataset.

Simulation and analysis have shown that selective search can reduce the cost of large-scale distributed information retrieval. By partitioning the collection into small topical shards, and then using a resource ranking algorithm to choose a subset of shards to search for each query, fewer postings are evaluated. Here we extend the study of selective search using a fine-grained simulation investigating: selective search efficiency in a parallel query processing environment; the difference in efficiency when term-based and sample-based resource selection algorithms are used; and the effect of two policies for assigning index shards to machines. Results obtained for two large datasets and four large query logs confirm that selective search is significantly more efficient than conventional distributed search. In particular, we show that selective search is capable of both higher throughput and lower latency in a parallel environment than is exhaustive search.

In Twitter, and other microblogging services, the generation of new content by the crowd is often biased towards immediacy: what is happening now. Prompted by the propagation of commentary and information through multiple mediums, users on the Web interact with and produce new posts about newsworthy topics and give rise to trending topics. This paper proposes to leverage on the behavioral dynamics of users to estimate the most relevant time periods for a topic. Our hypothesis stems from the fact that when a real-world event occurs it usually has peak times on the Web: a higher volume of tweets, new visits and edits to related Wikipedia articles, and news published about the event.

This paper presents EsdRank, a new technique for improving ranking using external semi-structured data such as controlled vocabularies and knowledge bases. EsdRank treats vocabularies, terms and entities from external data, as objects connecting query and documents. Evidence used to link query to objects, and to rank documents are incorporated as features between query-object and object-document correspondingly. A latent listwise learning to rank algorithm, Latent-ListMLE, models the objects as latent space between query and documents, and learns how to handle all evidence in a unified procedure from document relevance judgments. EsdRank is tested in two scenarios: Using a knowledge base for web search, and using a controlled vocabulary for medical search. Experiments on TREC Web Track and OHSUMED data show significant improvements over state-of-the-art baselines.

Selective distributed search is a retrieval architecture that reduces search costs by partitioning a corpus into topical shards such that only a few shards need to be searched for each query. Prior research created topical shards by using random seed documents to cluster a random sample of the full corpus. The resource selection algorithm might use a different random sample of the corpus. These random components make selective search non-deterministic. This paper studies how these random components affect experimental results. Experiments on two ClueWeb09 corpora and four query sets show that in spite of random components, selective search is stable for most queries.

Term weighting is a fundamental problem in IR research and numerous weighting models have been proposed. Proper term weighting can greatly improve retrieval accuracies, which essentially involves two types of query understanding: interpreting the query and judging the relative contribution of the terms to the query. These two steps are often dealt with separately, and complicated yet not so effective weighting strategies are proposed. In this paper, we propose to address query interpretation and term weighting in a unified framework built upon distributed representations of words from recent advances in neural network language modeling. Specifically, we represent term and query as vectors in the same latent space, construct features for terms using their word vectors and learn a model to map the features onto the defined target term weights. The proposed method is simple yet effective. Experiments using four collections and two retrieval models demonstrates significantly higher retrieval accuracies than baseline models.

The traditional search solution for large collections divides the collection into subsets ( shards ), and processes the query against all shards in parallel ( exhaustive search ). The search cost and the computational requirements of this approach are often prohibitively high for organizations with few computational resources. This article investigates and extends an alternative: selective search , an approach that partitions the dataset based on document similarity to obtain topic-based shards , and searches only a few shards that are estimated to contain relevant documents for the query. We propose shard creation techniques that are scalable, efficient, self-reliant, and create topic-based shards with low variance in size, and high density of relevant documents. The experimental results demonstrate that the effectiveness of selective search is on par with that of exhaustive search, and the corresponding search costs are substantially lower with the former. Also, the majority of the queries perform as well or better with selective search. An oracle experiment that uses optimal shard ranking for a query indicates that selective search can outperform the effectiveness of exhaustive search. Comparison with a query optimization technique shows higher improvements in efficiency with selective search. The overall best efficiency is achieved when the two techniques are combined in an optimized selective search approach.

Researchers have begun studying content obtained from microblogging services such as Twitter to address a variety of technological, social, and commercial research questions. The large number of Twitter users and even larger volume of tweets often make it impractical to collect and maintain a complete record of activity; therefore, most research and some commercial software applications rely on samples, often relatively small samples, of Twitter data. For the most part, sample sizes have been based on availability and practical considerations. Relatively little attention has been paid to how well these samples represent the underlying stream of Twitter data. To fill this gap, this article performs a comparative analysis on samples obtained from two of Twitter’s streaming APIs with a more complete Twitter dataset to gain an in-depth understanding of the nature of Twitter data samples and their potential for use in various data mining tasks.

State-of-the-art question answering (QA) systems employ passage retrieval based on bag-of-words similarity models with respect to a query and a passage. We propose a combination of a traditional bag-of-words similarity model and an annotation similarity model to improve passage ranking. The proposed annotation similarity model is generic enough to process annotations of arbitrary types. Historical fact validation is a subtask to determine whether a given sentence tells us historically correct information, which is important for a QA task on world history. Experimental results show that the combined model gains up to 7.7% and 4.2% improvements in historical fact validation in terms of precision at rank 1 and mean reciprocal rank, respectively.

Data retrieved from community question answering (CQA) sites, such as content and users' assessments of content, is commonly used for expertise estimation related tasks. One such task, in which the received votes are directly used as graded relevance assessment values, is ranking replies of a question. Even though these available assessments values are very practical for evaluation purposes, they may not always reflect the correct assessment value of the content, due to the possible temporal or presentation bias introduced by the CQA system during voting process. This paper analyzes a very commonly used CQA data collection in terms of these introduced biases and their effects on the experimental evaluation of approaches. A more bias free test set construction approach, which has correlated results with the manual assessments, is also proposed in this paper.

Large document collections can be partitioned into 'topical shards' to facilitate distributed search. In a low-resource search environment only a few of the shards can be searched in parallel. Such a search environment faces two intertwined challenges. First, determining which shards to consult for a given query: shard ranking. Second, how many shards to consult from the ranking: cutoff estimation. In this paper we present a family of three algorithms that address both of these problems. As a basis we employ a commonly used data structure, the central sample index (CSI), to represent the shard contents. Running a query against the CSI yields a flat document ranking that each of our algorithms transforms into a tree structure. A bottom up traversal of the tree is used to infer a ranking of shards and also to estimate a stopping point in this ranking that yields cost-effective selective distributed search. As compared to a state-of-the-art shard ranking approach the proposed algorithms provide substantially higher search efficiency while providing comparable search effectiveness.

People are seldom aware that their search queries frequently mismatch a majority of the relevant documents. This may not be a big problem for topics with a large and diverse set of relevant documents, but would largely increase the chance of search failure for less popular search needs. We aim to address the mismatch problem by developing accurate and simple queries that require minimal effort to construct. This is achieved by targeting retrieval interventions at the query terms that are likely to mismatch relevant documents. For a given topic, the proportion of relevant documents that do not contain a term measures the probability for the term to mismatch relevant documents, or the term mismatch probability. Recent research demonstrates that this probability can be estimated reliably prior to retrieval. Typically, it is used in probabilistic retrieval models to provide query dependent term weights. This paper develops a new use: Automatic diagnosis of term mismatch. A search engine can use the diagnosis to suggest manual query reformulation, guide interactive query expansion, guide automatic query expansion, or motivate other responses. The research described here uses the diagnosis to guide interactive query expansion, and create Boolean conjunctive normal form (CNF) structured queries that selectively expand 'problem' query terms while leaving the rest of the query untouched. Experiments with TREC Ad-hoc and Legal Track datasets demonstrate that with high quality manual expansion, this diagnostic approach can reduce user effort by 33%, and produce simple and effective structured queries that surpass their bag of word counterparts.

We describe a open-domain information extraction method for extracting concept-instance pairs from an HTML corpus. Most earlier approaches to this problem rely on combining clusters of distributionally similar terms and concept-instance pairs obtained with Hearst patterns. In contrast, our method relies on a novel approach for clustering terms found in HTML tables, and then assigning concept names to these clusters using Hearst patterns. The method can be efficiently applied to a large corpus, and experimental results on several datasets show that our method can accurately extract large numbers of concept-instance pairs.

editorial Free Access Share on TOIS Reviewers: October 2009 To September 2012 Editor: Jamie Callan View Profile Authors Info & Claims ACM Transactions on Information SystemsVolume 30Issue 4November 2012 Article No.: 27pp 1–3https://doi.org/10.1145/2382438.2382446Published:01 November 2012Publication History 0citation337DownloadsMetricsTotal Citations0Total Downloads337Last 12 Months21Last 6 weeks5 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteeReaderPDF

Aggregated search is the task of integrating results from potentially multiple specialized search services, or verticals, into the Web search results. The task requires predicting not only which verticals to present (the focus of most prior research), but also predicting where in the Web results to present them (i.e., above or below the Web results, or somewhere in between). Learning models to aggregate results from multiple verticals is associated with two major challenges. First, because verticals retrieve different types of results and address different search tasks, results from different verticals are associated with different types of predictive evidence (or features). Second, even when a feature is common across verticals, its predictiveness may be vertical-specific. Therefore, approaches to aggregating vertical results require handling an inconsistent feature representation across verticals, and, potentially, a vertical-specific relationship between features and relevance. We present 3 general approaches that address these challenges in different ways and compare their results across a set of 13 verticals and 1070 queries. We show that the best approaches are those that allow the learning algorithm to learn a vertical-specific relationship between features and relevance.

Question-answering, computer-assisted language learning, text mining, and other software applications that use a full-search engine to find information in a large text corpus are becoming common. A software application may use metadata and text annotations to reduce the mismatch between the concept-based representations convenient for inference and the word-based representations typically used for text retrieval. Software applications may also be able to specify detailed requirements that retrieved passages must satisfy. This use of text search is very different than the ad-hoc, interactive search that information retrieval research typically studies.

Indexes for large collections are often divided into shards that are distributed across multiple computers and searched in parallel to provide rapid interactive search. Typically, all index shards are searched for each query. For organizations with modest computational resources the high query processing cost incurred in this exhaustive search setup can be a deterrent to working with large collections. This paper investigates document allocation policies that permit searching only a few shards for each query (selective search) without sacrificing search accuracy. Random, source-based and topic-based document-to-shard allocation policies are studied in the context of selective search.

The probability that a term appears in relevant documents (P(t | R)) is a fundamental quantity in several probabilistic retrieval models, however it is difficult to estimate without relevance judgments or a relevance model. We call this value term necessity because it measures the percentage of relevant documents retrieved by the term - how necessary a term's occurrence is to document relevance. Prior research typically either set this probability to a constant, or estimated it based on the term's inverse document frequency, neither of which was very effective.

In some retrieval situations, a system must search across multiple collections. This task, referred to as federated search, occurs for example when searching a distributed index or aggregating content for web search. Resource selection refers to the subtask of deciding, given a query, which collections to search. Most existing resource selection methods rely on evidence found in collection content. We present an approach to resource selection that combines multiple sources of evidence to inform the selection decision. We derive evidence from three different sources: collection documents, the topic of the query, and query click-through data. We combine this evidence by treating resource selection as a multiclass machine learning problem. Although machine learned approaches often require large amounts of manually generated training data, we present a method for using automatically generated training data. We make use of and compare against prior resource selection work and evaluate across three experimental testbeds.

Search engines that support structured documents typically support structure created by the author (e.g., title, section), and may also support structure added by an annotation process (e.g., part of speech, named entity, semantic role). Exploiting such structure can be difficult. Query structure may fail to match structure in a relevant document for a variety of reasons, thus structured queries, although containing more information than keyword queries, are often less effective than unstructured queries. This paper studies retrieval of sentences with annotations for a question answering task. Three problems of structured retrieval are identified and solutions proposed. Structural mismatch is addressed by query structure expansion of predicted relevant structures. Lack of presence of all key aspects of a question is solved by Boolean filtering of result sentences. The score variations of the annotator generated fields with all the different lengths are accounted for by using field specific smoothing. Experiments show that each solution incrementally improves structured retrieval, and a combination of Boolean filtering, structural expansion, and keyword queries outperforms keyword and simple structured retrieval baselines.

Web search providers often include search services for domain-specific subcollections, called verticals, such as news, images, videos, job postings, company summaries, and artist profiles. We address the problem of vertical selection, predicting relevant verticals (if any) for queries issued to the search engine's main web search page. In contrast to prior query classification and resource selection tasks, vertical selection is associated with unique resources that can inform the classification decision. We focus on three sources of evidence: (1) the query string, from which features are derived independent of external resources, (2) logs of queries previously issued directly to the vertical, and (3) corpora representative of vertical content. We focus on 18 different verticals, which differ in terms of semantics, media type, size, and level of query traffic. We compare our method to prior work in federated search and retrieval effectiveness prediction. An in-depth error analysis reveals unique challenges across different verticals and provides insight into vertical selection for future work.

Most existing automatic taxonomy induction systems exploit one or more features to induce a taxonomy; nevertheless there is no systematic study examining which are the best features for the task under various conditions. This paper studies the impact of using different features on taxonomy induction for different types of relations and for terms at different abstraction levels. The evaluation shows that different conditions need different technologies or different combination of the technologies. In particular, co-occurrence and lexico-syntactic patterns are good features for is-a, sibling and part-of relations; contextual, co-occurrence, patterns, and syntactic features work well for concrete terms; co-occurrence works well for abstract terms.

Structured documents contain elements defined by the author(s) and annotations assigned by other people or processes. Structured documents pose challenges for probabilistic retrieval models when there are mismatches between the structured query and the actual structure in a relevant document or erroneous structure introduced by an annotator. This paper makes three contributions. First, a new generative retrieval model is proposed to deal with the mismatch problem. This new model extends the basic keyword language model by treating structure as hidden variable during the generation process. Second, variations of the model are compared. Third, term-level and structure-level smoothing strategies are studied. Evaluation was conducted with INEX XML retrieval and question-answering retrieval tasks. Experimental results indicate that the optimal structured retrieval model is task dependent, two-level Dirichlet smoothing significantly outperforms two-level Jelinek-Mercer smoothing, and with accurate structured queries, the proposed structured retrieval model outperforms keyword retrieval significantly, on both QA and INEX datasets.

Blog feed search poses different and interesting challenges from traditional ad hoc document retrieval. The units of retrieval, the blogs, are collections of documents, the blog posts. In this work we adapt a state-of-the-art federated search model to the feed retrieval task, showing a significant improvement over algorithms based on the best performing submissions in the TREC 2007 Blog Distillation task[12]. We also show that typical query expansion techniques such as pseudo-relevance feedback using the blog corpus do not provide any significant performance improvement and in many cases dramatically hurt performance. We perform an in-depth analysis of the behavior of pseudo-relevance feedback for this task and develop a novel query expansion technique using the link structure in Wikipedia. This query expansion technique provides significant and consistent performance improvements for this task, yielding a 22% and 14% improvement in MAP over the unexpanded query for our baseline and federated algorithms respectively.

Personalized search is a promising way to better serve different users' information needs. Search history is one of the major information sources for search personalization. We investigated the impact of history length on the effectiveness of personalized ranking. We carried out task-based user study for Web search, and obtained ranked relevance judgments for all queries. Query contexts derived from previous queries in the same task are used to re-rank results for the current query. Experimental results show that the performance of personalization generally improves as more queries are accumulated, but most of the benefits come from a few immediately preceding queries.

Bag-of-words retrieval is popular among Question Answering (QA) system developers, but it does not support constraint checking and ranking on the linguistic and semantic information of interest to the QA system. We present anapproach to retrieval for QA, applying structured retrieval techniques to the types of text annotations that QA systems use. We demonstrate that the structured approach can retrieve more relevant results, more highly ranked, compared with bag-of-words, on a sentence retrieval task. We also characterize the extent to which structured retrieval effectiveness depends on the quality of the annotations.

Existing pseudo-relevance feedback methods typically perform averaging over the top-retrieved documents, but ignore an important statistical dimension: the risk or variance associated with either the individual document models, or their combination. Treating the baseline feedback method as a black box, and the output feedback model as a random variable, we estimate a posterior distribution for the feed-back model by resampling a given query's top-retrieved documents, using the posterior mean or mode as the enhanced feedback model. We then perform model combination over several enhanced models, each based on a slightly modified query sampled from the original query. We find that resampling documents helps increase individual feedback model precision by removing noise terms, while sampling from the query improves robustness (worst-case performance) by emphasizing terms related to multiple query aspects. The result is a meta-feedback algorithm that is both more robust and more precise than the original strong baseline method.

A version of cobweb/classit is proposed to incrementally cluster text documents into cluster hierarchies. The modification to classit consists of changes to the underlying distributional assumption of the original algorithm that are suggested by text document data. Both the algorithms are evaluated using standard text document datasets. We show that the modified algorithm performs better than the original Classit when presented with Reuters newswire articles in temporal order, i.e., the order in which they are going to be presented in real life situation. It also performs better than the original Classit on the larger of eleven standard text clustering datasets we used.

User modeling for information retrieval has mostly been studied to improve the effectiveness of information access in centralized repositories. In this paper we explore user modeling in the context of full-text federated search in peer-to-peer networks. Our approach models a user's persistent, long-term interests based on past queries, and uses the model to improve search efficiency for future queries that represent interests similar to past queries. Our approach also enables queries representing a user's transient, ad-hoc interests to be automatically recognized so that search for these queries can rely on a relatively large search radius to avoid sacrificing effectiveness for efficiency. Experimental results demonstrate that our approach can significantly improve the efficiency of full-text federated search without degrading its accuracy. Furthermore, the proposed approach does not require a large amount of training data, and is robust to a range of parameter values.

Document hierarchies provide views of a collection at different levels of granularity, making it easy to visualize and explore large document collections. Topic descriptors at each level of the hierarchy play an important role in helping users to achieve those

For the task of near-duplicated document detection, both traditional fingerprinting techniques used in database community and bag-of-word comparison approaches used in information retrieval community are not sufficiently accurate. This is due to the fact that the characteristics of near-duplicated documents are different from that of both “almost-identical ” documents in the data cleaning task and “relevant ” documents in the search task. This paper presents an instance-level constrained clustering approach for near-duplicate detection. The framework incorporates information such as document attributes and content structure into the clustering process to form near-duplicate clusters. Gathered from several collections of public comments sent to U.S. government agencies on proposed new regulations, the experimental results demonstrate that our approach outperforms other near-duplicate detection algorithms and as about as effective as human assessors.

Abstract Federated search and distributed information retrieval systems provide a single user interface for searching multiple full‐text search engines. They have been an active area of research for more than a decade, but in spite of their success as a research topic, they are still rare in operational environments. This article discusses a prototype federated search system developed for the U.S. government's FedStats Web portal, and the issues addressed in adapting research solutions to this operational environment. A series of experiments explore how well prior research results, parameter settings, and heuristics apply in the FedStats environment. The article concludes with a set of lessons learned from this technology transfer effort, including observations about search engine quality in the “real world.”

It has long been recognized that capturing term relationships is an important aspect of information retrieval. Even with large amounts of data, we usually only have significant evidence for a fraction of all potential term pairs. It is therefore important to consider whether multiple sources of evidence may be combined to predict term relations more accurately. This is particularly important when trying to predict the probability of relevance of a set of terms given a query, which may involve both lexical and semantic relations between the terms.We describe a Markov chain framework that combines multiple sources of knowledge on term associations. The stationary distribution of the model is used to obtain probability estimates that a potential expansion term reflects aspects of the original query. We use this model for query expansion and evaluate the effectiveness of the model by examining the accuracy and robustness of the expansion methods, and investigate the relative effectiveness of various sources of term evidence. Statistically significant differences in accuracy were observed depending on the weighting of evidence in the random walk. For example, using co-occurrence data later in the walk was generally better than using it early, suggesting further improvements in effectiveness may be possible by learning walk behaviors.

Federated search links multiple search engines into a single, virtual search system. Most prior research of federated search focused on selecting search engines that have the most relevant contents, but ignored the retrieval effectiveness of individual search engines. This omission can cause serious problems when federating search engines of different qualities.This paper proposes a federated search technique that uses utility maximization to model the retrieval effectiveness of each search engine in a federated search environment. The new algorithm ranks the available resources by explicitly estimating the amount of relevant material that each resource can return, instead of the amount of relevant material that each resource contains. An extensive set of experiments demonstrates the effectiveness of the new algorithm.

Abstract A potentially useful feature of information retrieval systems for students is the ability to identify documents that not only are relevant to the query but also match the student's reading level. Manually obtaining an estimate of reading difficulty for each document is not feasible for very large collections, so we require an automated technique. Traditional readability measures, such as the widely used Flesch‐Kincaid measure, are simple to apply but perform poorly on Web pages and other nontraditional documents. This work focuses on building a broadly applicable statistical model of text for different reading levels that works for a wide range of documents. To do this, we recast the well‐studied problem of readability in terms of text categorization and use straightforward techniques from statistical language modeling. We show that with a modified form of text categorization, it is possible to build generally applicable classifiers with relatively little training data. We apply this method to the problem of classifying Web pages according to their reading difficulty level and show that by using a mixture model to interpolate evidence of a word's frequency across grades, it is possible to build a classifier that achieves an average root mean squared error of between one and two grade levels for 9 of 12 grades. Such classifiers have very efficient implementations and can be applied in many different scenarios. The models can be varied to focus on smaller or larger grade ranges or easily retrained for a variety of tasks or populations.

Hierarchical peer to peer networks with multiple directory services are an important architecture for large-scale file sharing due to their effectiveness and efficiency. Recent research argues that they are also an effective method of providing large-scale content-based federated search of text-based digital libraries. In both cases the directory services are critical resources that are subject to attack or failure, but the latter architecture may be particularly vulnerable because content is less likely to be replicated throughout the network.

This paper presents a unified utility framework for resource selection of distributed text information retrieval. This new framework shows an efficient and effective way to infer the probabilities of relevance of all the documents across the text databases. With the estimated relevance information, resource selection can be made by explicitly optimizing the goals of different applications. Specifically, when used for database recommendation, the selection is optimized for the goal of high-recall (include as many relevant documents as possible in the selected databases); when used for distributed document retrieval, the selection targets the high-precision goal (high precision in the final merged list of documents). This new model provides a more solid framework for distributed information retrieval. Empirical studies show that it is at least as effective as other state-of-the-art algorithms.

No abstract available.

INTRODUCTION A widely-used architecture for factoid question answering (QA) involves the use of a multi-step pipeline consisting of: 1) initial question analysis, 2) document and/or passage retrieval, and 3) answer extraction. In this study, we examine the relationship between the quality of document retrieval and the overall accuracy of QA systems. We evaluate two QA systems using TREC 2002 test set questions [9]: Carnegie Mellon's JAVELIN system [7] and Waterloo's MultiText QA system [2]. We adapt the two QA systems in order to use di#erent sets of documents as input, and seven different document retrieval methods to create the list of documents including a combination of di#erent systems. The set of known relevant documents was used as a baseline to compare the di#erent retrieval methods. Documents with exact or inexact judgments are considered relevant. Our main hypothesis for this study is that there is a positive relationship between improved document retrieval and QA accuracy

No abstract available.

No abstract available.

In this paper, we describe a new model for collaborative filtering. The motivation of this work comes from the fact that two users with very similar preferences on items may have very different rating schemes. For example, one user may tend to assign a higher rating to all items than another user. Unlike previous models of collaborative filtering, which determine the similarity between two users only based on their rating performance, our model treats the user's preferences on items separately from the user's rating scheme. More specifically, for each user, we build two separate models: a preference model capturing which items are favored by the user and a rating model capturing how the user would rate an item given the preference information. The similarity of two users is computed based on the underlying preference model, instead of the surface ratings. We compare the new model with several representative previous approaches on two data sets. Experiment results show that the new model outperforms all the previous approaches that are tested consistently on both data sets.

Hybrid peer-to-peer architectures use special nodes to provide directory services for regions of the network ("regional directory services"). Hybrid peer-to-peer architectures are a potentially powerful model for developing large-scale networks of complex digital libraries, but peer-to-peer networks have so far tended to use very simple methods of resource selection and document retrieval. In this paper, we study the application of content-based resource selection and document retrieval to hybrid peer-to-peer networks. The directory nodes that provide regional directory services construct and use the content models of neighboring nodes to determine how to route query messages through the network. The leaf nodes that provide information use content-based retrieval to decide which documents to retrieve for queries. The experimental results demonstrate that using content-based retrieval in hybrid peer-to-peer networks is both more accurate and more efficient for some digital library environments than more common alternatives such as Gnutella 0.6.

No abstract available.

Callan, Jamie (ed.): SIGIR '03: Proceedings of the 26th annual international ACM SIGIR conference on Research and development in informaion retrieval. New York: ACM, 2003, 363–364. - ISBN: 978-1-58113-646-3 Online also available at: https://doi.org/10.1145/860435.860502

This paper investigates the pre-conditions for successful combination of document representations formed from structural markup for the task of known-item search. As this task is very similar to work in meta-search and data fusion, we adapt several hypotheses from those research areas and investigate them in this context. To investigate these hypotheses, we present a mixture-based language model and also examine many of the current meta-search algorithms. We find that compatible output from systems is important for successful combination of document representations. We also demonstrate that combining low performing document representations can improve performance, but not consistently. We find that the techniques best suited for this task are robust to the inclusion of poorly performing document representations. We also explore the role of variance of results across systems and its impact on the performance of fusion, with the surprising result that the correct documents have higher variance across document representations than highly ranking incorrect documents.

Prior research under a variety of conditions has shown the CORI algorithm to be one of the most effective resource selection algorithms, but the range of database sizes studied was not large. This paper shows that the CORI algorithm does not do well in environments with a mix of "small" and "very large" databases. A new resource selection algorithm is proposed that uses information about database sizes as well as database contents. We also show how to acquire database size estimates in uncooperative environments as an extension of the query-based sampling used to acquire resource descriptions. Experiments demonstrate that the database size estimates are more accurate for large databases than estimates produced by a competing method; the new resource ranking algorithm is always at least as effective as the CORI algorithm; and the new algorithm results in better document rankings than the CORI algorithm.

The proliferation of searchable text databases on local area networks and the Internet causes the problem of finding information that may be distributed among many disjoint text databases ( distributed information retrieval ). How to merge the results returned by selected databases is an important subproblem of the distributed information retrieval task. Previous research assumed that either resource providers cooperate to provide normalizing statistics or search clients download all retrieved documents and compute normalized scores without cooperation from resource providers.This article presents a semisupervised learning solution to the result merging problem. The key contribution is the observation that information used to create resource descriptions for resource selection can also be used to create a centralized sample database to guide the normalization of document scores returned by different databases. At retrieval time, the query is sent to the selected databases, which return database-specific document scores, and to a centralized sample database , which returns database-independent document scores. Documents that have both a database-specific score and a database-independent score serve as training data for learning to normalize the scores of other documents. An extensive set of experiments demonstrates that this method is more effective than the well-known CORI result-merging algorithm under a variety of conditions.

The usual approach to named-entity detection is to learn extraction rules that rely on linguistic, syntactic, or document format patterns that are consistent across a set of documents. However, when there is no consistency among documents, it may be more effective to learn document-specific extraction rules.This paper presents a knowledge-based approach to learning rules for named-entity extraction. Document-specific extraction rules are created using a generate-and-test paradigm and a database of known named-entities. Experimental results show that this approach is effective on Web documents that are difficult for the usual methods.

Query-based sampling is a method of discovering the contents of a text database by submitting queries to a search engine and observing the documents returned. In prior research sampled documents were used to build resource descriptions for automatic database selection, and to build a centralized sample database for query expansion and result merging. An unstated assumption was that the associated storage costs were acceptable.When sampled documents are long, storage costs can be large. This paper investigates methods of pruning long documents to reduce storage costs. The experimental results demonstrate that building resource descriptions and centralized sample databases from the pruned contents of sampled documents can reduce storage costs by 54-93% while causing only minor losses in the accuracy of distributed information retrieval.

Statistical language models have been proposed recently for several information retrieval tasks, including the resource selection task in distributed information retrieval. This paper extends the language modeling approach to integrate resource selection, ad-hoc searching, and merging of results from different text databases into a single probabilistic retrieval model. This new approach is designed primarily for Intranet environments, where it is reasonable to assume that resource providers are relatively homogeneous and can adopt the same kind of search engine. Experiments demonstrate that this new, integrated approach is at least as effective as the prior state-of-the-art in distributed IR.

Information retrieval using meta data can be traced back to the early age of IR where documents are represented by the controlled vocabulary. In this paper, we explore the usage of meta-data information under the framework of language model. We present a new language model that is able to take advantage of the category information for documents to improve the retrieval accuracy. We compare the new language model with the traditional language model over the TREC4 dataset where the collection information for documents is obtained using the k-means clustering method. The new language model outperforms the traditional language model, which verifies our statement.

This paper addresses the problem of merging results obtained from different databases and search engines in a distributed information retrieval environment. The prior research on this problem either assumed the exchange of statistics necessary for normalizing scores (cooperative solutions) or is heuristic. Both approaches have disadvantages. We show that the problem in uncooperative environments is simpler when viewed as a component of a distributed IR system that uses query-based sampling to create resource descriptions. Documents sampled for creating resource descriptions can also be used to create a sample centralized index, and this index is a source of training data for adaptive results merging algorithms. A variety of experiments demonstrate that this new approach is more effective than a well-known alternative, and that it allows query-by-query tuning of the results merging function.

This paper addresses the problem of extending an adaptive information filtering system to make decisions about the novelty and redundancy of relevant documents. It argues that relevance and redundance should each be modelled explicitly and separately. A set of five redundancy measures are proposed and evaluated in experiments with and without redundancy thresholds. The experimental results demonstrate that the cosine similarity metric and a redundancy measure based on a mixture of language models are both effective for identifying redundant documents.

Query expansion has been shown effective for both single database retrieval and for distributed information retrieval where complete collection information is available. One might expect that query expansion would then work for distributed information retrieval when complete collection information is not available. However, this does not appear to be the case. When using local context analysis for query expansion in distributed retrieval with partial information, the most significant reason query expansion does not work is that merging scores of documents retrieved by expanded queries is very difficult. However, we have found that using sampled information for query expansion can give boosts in a single database environment, and that when more information is available, query expansion can work in distributed environments. We also show that most of the benefit of query expansion in distributed retrieval comes from finding good documents, and not from selecting good databases. 1

In this paper, we present a new method of using statistical models to estimate readability [1]. Language Model is used to capture the content information. It is combined with linguistic feature model by a linear form. Experiments show that this new method has a better performance than the widely used Flesch-Kincaid readability formula.

Information filtering systems based on statistical retrieval models usually compute a numeric score indicating how well each document matches each profile. Documents with scores above profile-specificdissemination thresholdsare delivered.

The proliferation of searchable text databases on corporate networks and the Internet causes a database selection problem for many people. Algorithms such as gGLOSS and CORI can automatically select which text databases to search for a given information need, but only if given a set of resource descriptions that accurately represent the contents of each database. The existing techniques for a acquiring resource descriptions have significant limitations when used in wide-area networks controlled by many parties. This paper presents query-based sampling , a new technicque for acquiring accurate resource descriptions. Query-based sampling does not require the cooperation of resource providers, nor does it require that resource providers use a particular search engine or representation technique. An extensive set of experimental results demonstrates that accurate resource descriptions are crated, that computation and communication costs are reasonable, and that the resource descriptions do in fact enable accurate automatic dtabase selection.

This paper discusses passage extraction approaches to multidocument summarization that use available information about the document set as a whole and the relationships between the documents to build on single document summarization methodology.Multi-document summarization diers from single in that the issues of compression, speed, redundancy and passage selection are critical in the formation of useful summaries, as well as the user's goals in creating the summary.Our approach addresses these issues by using domain-independent techniques based mainly on fast, statistical processing, a metric for reducing redundancy and maximizing diversity in the selected passages, and a modular framework to allow easy parameterization for dierent genres, corpora characteristics and user requirements.We examined how h umans create multi-document summaries as well as the characteristics of such summaries and use these summaries to evaluate the performance of various multidocument summarization algorithms.

We investigate three issues in distributed information retrieval, considering both TREC data and U.S. Patents: (1) topical organization of large text collections, (2) collection ranking and selection with topically organized collections (3) results merging, particularly document score normalization, with topically organized collections. We find that it is better to organize collections topically, and that topical collections can be well ranked using either INQUERY's CORI algorithm, or the Kullback-Leibler divergence (KL), but KL is far worse than CORI for non-topically organized collections. For results merging, collections organized by topic require global idfs for the best performance. Contrary to results found elsewhere, normalized scores are not as good as global idfs for merging when the collections are topically organized. Keywords Information retrieval, Collection selection, Topical organization. 1. INTRODUCTION We have developed a distributed system for the search and class...

Abstract The proliferation of online information resources increases the importance of effective and efficient distributed searching. Distributed searching is cast in three parts – database selection, query processing, and results merging. In this paper we examine the effect of database selection on retrieval performance. We look at retrieval performance in three different distributed retrieval testbeds and distill some general results. First we find that good database selection can result in better retrieval effectiveness than can be achieved in a centralized database. Second we find that good performance can be achieved when only a few sites are selected and that the performance generally increases as more sites are selected. Finally we find that when database selection is employed, it is not necessary to maintain collection wide information (CWI), e.g. global idf. Local information can be used to achieve superior performance. This means that distributed systems can be engineered with more autonomy and less cooperation. This work suggests that improvements in database selection can lead to broader improvements in retrieval performance, even in centralized (i.e. single database) systems. Given a centralized database and a good selection mechanism, retrieval performance can be improved by decomposing that database conceptually and employing a selection step. 1

We compare the performance of two database selection algorithms reported in the literature.Their performance is compared using a common testbed designed specifically for database selection techniques.The testbed is a decomposition of the TREC/TIPSTER data into 236 subcollections.The databases from our testbed were ranked using both the gGlOSS and CORI techniques and compared to a baseline derived from TREC relevance judgements.We examined the degree to which CORI and gGlOSS approximate this baseline.Our results con rm our earlier observation that the gGlOSS Ideal(l) ranks do not estimate relevancebased ranks well.We also nd that CORI is a uniformly better estimator of relevance-based ranks than gGlOSS for the test environment used in this study.Part of the advantage of the CORI algorithm can be explained by a strong correlation between gGlOSS and a size-based baseline (SBR).We also nd that CORI produces consistently accurate rankings on testbeds ranging from 100{921 sites.However for a given level of recall, search eort appears to scale linearly with the number of databases.

This paper examines the problems of learning queries and dissemination thresholds from relevance feedback in a dynamic information filtering environment.It revisits the EG algorithm for learning queries, identifying several problems in using it reliably for information filtering, and providing solutions.It also presents a new algorithm for learning dissemination thresholds automatically, from the same relevance feedback information used to learn queries.

This paper evaluates the retrieval effectiveness of distributed information retrieval systems in realistic environments. We find that when a large number of collections are available, the retrieval effectiveness is significantly worse than that of centralized systems, mainly because typical queries are not adequate for the purpose of choosing the right collections. We propose two techniques to address the problem. One is to use phrase information in the collection selection index and the other is query expansion. Both techniques enhance the discriminatory power of typical queries for choosing the right collections and hence significantly improve retrieval results. Query expansion, in particular, brings the effectiveness of searching a large set of distributed collections close to that of searching a centralized collection. 1 Introduction In today's network environments, information is highly distributed. The Internet or World Wide Web, for example, contains thousands of collections. ...

Article Document filtering with inference networks Share on Author: Jamie Callan Computer Science Department, University of Massachusetts, Amherst, MA Computer Science Department, University of Massachusetts, Amherst, MAView Profile Authors Info & Claims SIGIR '96: Proceedings of the 19th annual international ACM SIGIR conference on Research and development in information retrievalAugust 1996 Pages 262–269https://doi.org/10.1145/243199.243273Online:18 August 1996Publication History 64citation457DownloadsMetricsTotal Citations64Total Downloads457Last 12 Months3Last 6 weeks0 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

Systems for text retrieval, routing, categorization and other IR tasks rely heavily on linear classifiers.We propose that two machine learning algorithms, the Widrow-Hoff and EG algorithms, be used in training linear text classifiers.In contrast to most IR methods, theoretical analysis provides performance guarantees and guidance on parameter settings for these algorithms.Experimental data is presented showing Widrow-Hoff and EG to be more effective than the widely used Rocchio algorithm on several categorization and routing tasks.

The use of information retrieval systems in networked environments raises a new set of issues that have received little attention.These issues include ranking document collections for relevance to a query, selecting the best set of collections from a ranked list, and merging the document rankings that are returned from a set of collections.This paper describes methods of addressing each issue in the inference network model, dkcusses their implementation in the INQUERY system, and presents experimental results demonstrating their effectiveness.1

The full integration of information retrieval (IR) features into a database management system (DBMS) has long been recognized as both a significant goal and a challenging undertaking. By full integration we mean: i) support for document storage, indexing, retrieval, and update, ii) transaction semantics, thus all database operations on documents have the ACID properties of atomicity, consistency, isolation, and durability, iii) concurrent addition, update, and retrieval of documents, and iv) database query language extensions to provide ranking for document retrieval operations. It is also necessary for the integrated offering to exhibit scaleable performance for document indexing and retrieval processes. To identify the implementation requirements imposed by the desired level of integration, we layered a representative IR application on Oracle Rdb and then conducted a number of database load and document retrieval experiments. The results of these experiments suggest that infrastructural extensions are necessary to obtain both the desired level of IR integration and scaleable performance. With the insight gained from our initial experiments, we developed an approach, called cooperative indexing, that provides a framework to achieve both scaleability and full integration of IR and RDBMS technology. Prototype implementations of system-level extensions to support cooperative indexing were evaluated with a modified version of Oracle Rdb. Our experimental findings validate the cooperative indexing scheme and suggest alternatives to further improve performance.

Five independently generated Boolean query formulations for ten different TREC topics were produced by ten different expert online searchers. These different formulations were grouped, and the groups, and combinations of them, were used as searches against the TREC test collection, using the INQUERY probabilistic inference network retrieval engine. Results show that progressive combination of query formulations leads to progressively improving retrieval performance. Results were compared against the performance of INQUERY natural language based queries, and in combination with them. The issue of recall as a performance measure in large databases was raised, since overlap between the searches conducted in this study, and the TREC-1 searches, was smaller than expected.

The TIPSTER collection is unusual because of both its size and detail. In particular, it describes a set of information needs, as opposed to traditional queries. These detailed representations of information need are an opportunity for research on different methods of formulating queries. This paper describes several methods of constructing queries for the INQUERY information retrieval system, and then evaluates those methods on the TIPSTER document collection. Both AdHoc and Routing query processing methods are evaluated.