User simulation is essential for generating enough data to train a statistical spoken dialogue system. Previous models for user simulation suffer from several drawbacks, such as the inability to take dialogue history into account, the need of rigid structure to ensure coherent user behaviour, heavy dependence on a specific domain, the inability to output several user intentions during one dialogue turn, or the requirement of a summarized action space for tractability. This paper introduces a data-driven user simulator based on an encoder-decoder recurrent neural network. The model takes as input a sequence of dialogue contexts and outputs a sequence of dialogue acts corresponding to user intentions. The dialogue contexts include information about the machine acts and the status of the user goal. We show on the Dialogue State Tracking Challenge 2 (DSTC2) dataset that the sequence-to-sequence model outperforms an agenda-based simulator and an n-gram simulator, according to F-score. Furthermore, we show how this model can be used on the original action space and thereby models user behaviour with finer granularity.

In this paper, we propose to use deep policy networks which are trained with an advantage actor-critic method for statistically optimised dialogue systems. First, we show that, on summary state and action spaces, deep Reinforcement Learning (RL) outperforms Gaussian Processes methods. Summary state and action spaces lead to good performance but require pre-engineering effort, RL knowledge, and domain expertise. In order to remove the need to define such summary spaces, we show that deep RL can also be trained efficiently on the original state and action spaces. Dialogue systems based on partially observable Markov decision processes are known to require many dialogues to train, which makes them unappealing for practical deployment. We show that a deep RL method based on an actor-critic architecture can exploit a small amount of data very efficiently. Indeed, with only a few hundred dialogues collected with a handcrafted policy, the actor-critic deep learner is considerably bootstrapped from a combination of supervised and batch RL. In addition, convergence to an optimal policy is significantly sped up compared to other deep RL methods initialized on the data with batch RL. All experiments are performed on a restaurant domain derived from the Dialogue State Tracking Challenge 2 (DSTC2) dataset.

Search result diversification (SRD) aims to select diverse documents from the search results in order to cover as many search intents as possible. A prerequisite is that the search results contain diverse documents. For this purpose, we investigate a new approach to SRD by diversifying the query. Expansion terms are selected from ConceptNet so as to cover as diverse aspects as possible. The experimental results on several TREC data sets show that our method can outperform the existing state-of-the-art approaches that do not diversify the query.

Recently, increasing attention has been given to a possible reinterpretation of information retrieval issues in the more general probabilistic framework offered by Quantum Theory.

Lots of document collections are well organized in hierarchical structure, and such structure can help users browse and understand these collections. Meanwhile, there are a large number of plain document collections loosely organized, and it is difficult for users to understand them effectively. In this paper we study how to automatically integrate latent topics in a plain collection with the topics in a hierarchical structured collection. We propose to use semi-supervised topic modeling to solve the problem in a principled way. The experiments show that the proposed method can generate both meaningful latent topics and expand high quality hierarchical topic structures.

Social annotations provide additional document description contributed by online users and they have been explored for improving search performance. However, most existing methods need offline analysis of the whole tagged corpus, which is computationally expensive and cannot fit specific queries well. In this paper, we propose to use tags for document re-ranking. Specifically, we first estimate document similarity by combining words and tags and then adjust the document ranks with the assumption that similar documents should have similar retrieval scores. On similarity estimation, we present a new feature extraction method, called CRMF, from which document similarity can be derived. The CRMF can integrate the content and relation properties of multiple views and mine their correspondence. Besides, it does not require that all the documents to have tags. We tested the proposed approach on collections which are derived from Clue Web and contain Delicious tags. The experimental results demonstrate the effectiveness of tags on document re-ranking, where CRMF is significantly better than other state-of-the-art methods using tags.

Large web search engines are facing formidable performance challenges because they have to process thousands of queries per second on tens of billions of documents, within interactive response time. Among many others, Top-k query processing (also called early termination or dynamic pruning) is an important class of optimization techniques that can improve the search efficiency and achieve faster query processing by avoiding the scoring of documents that are unlikely to be in the top results. One recent technique is using Block-Max index. In the Block-Max index, the posting lists are organized as blocks and the maximum score for each block is stored to improve the query efficiency. Although query processing speedup is achieved with Block-Max index, the ranking function for the Top-k results is the term-based approach. It is well known that documents' static scores are also important for a good ranking function. In this paper, we show that the performance of the state-of-the-art algorithms with the Block-Max index is degraded when the static score is added in the ranking function. Then we study efficient techniques for Top-k query processing in the case where a page's static score is given, such as PageRank, in addition to the term-based approach. In particular, we propose a set of new algorithms based on the WAND and MaxScore with Block-Max index using local score, which outperform the existing ones. Then we propose new techniques to estimate a better score upper bound for each block. We also study the search efficiency on different index structures where the document identifiers are assigned by URL sorting or by static document scores. Experiments on TREC GOV2 and ClueWeb09B show that considerable performance gains are achieved.

Citation Recommendation is useful for an author to find out the papers or books that can support the materials she is writing about. It is a challengeable problem since the vocabulary used in the content of papers and in the citation contexts are usually quite different. To address this problem, we propose to use translation model, which can bridge the gap between two heterogeneous languages. We conduct an experiment and find the translation model can provide much better candidates of citations than the state-of-the-art methods.

A large proportion of search engine queries contain phrases,namely a sequence of adjacent words. In this paper, we propose to use flat position index (a.k.a schema-independent index) for phrase query evaluation. In the flat position index, the entire document collection is viewed as a huge sequence of tokens. Each token is represented by one flat position, which is a unique position offset from the beginning of the collection. Each indexed term is associated with a list of the flat positions about that term in the sequence. To recover DocID from flat positions efficiently, we propose a novel cache sensitive look-up table (CSLT), which is much faster than existing search algorithms. Experiments on TREC GOV2 data collection show that flat position index can reduce the index size and speed up phrase querying substantially, compared with traditional word-level index.

Various click models have been recently proposed as a principled approach to infer the relevance of documents from the clickthrough data. The inferred document relevance is potentially useful in evaluating the Web retrieval systems. In practice, it generally requires to acquire the accurate evaluation results within minimal users' query submissions. This problem is important for speeding up search engine development and evaluation cycle and acquiring reliable evaluation results on tail queries. In this paper, we propose a reordering framework for efficient evaluation problem in the context of clickthrough based Web retrieval evaluation. The main idea is to move up the documents that contribute more for the evaluation task. In this framework, we propose four intuitions and formulate them as an optimization problem. Both user study and TREC data based experiments validate that the reordering framework results in much fewer query submissions to get accurate evaluation results with only a little harm to the users' utility.

Bursty features in text streams are very useful in many text mining applications. Most existing studies detect bursty features based purely on term frequency changes without taking into account the semantic contexts of terms, and as a result the detected bursty features may not always be interesting or easy to interpret. In this paper we propose to model the contexts of bursty features using a language modeling approach. We then propose a novel topic diversity-based metric using the context models to find newsworthy bursty features. We also propose to use the context models to automatically assign meaningful tags to bursty features. Using a large corpus of a stream of news articles, we quantitatively show that the proposed context language models for bursty features can effectively help rank bursty features based on their newsworthiness and to assign meaningful tags to annotate bursty features.

The Cranfield evaluation method has some disadvantages, including its high cost in labor and inadequacy for evaluating interactive retrieval techniques. As a very promising alternative, automatic comparison of retrieval systems based on observed clicking behavior of users has recently been studied. Several methods have been proposed, but there has so far been no systematic way to assess which strategy is better, making it difficult to choose a good method for real applications. In this paper, we propose a general way to evaluate these relative comparison methods with two measures: utility to users(UtU) and effectiveness of differentiation(EoD). We evaluate two state of the art methods by systematically simulating different retrieval scenarios. Inspired by the weakness of these methods revealed through our evaluation, we further propose a novel method by considering the positions of clicked documents. Experiment results show that our new method performs better than the existing methods.

Organizing Web search results into labeled categories is a difficult but very useful task. The idea is to group the many results that each user query generates into well-labeled categories, so that users can find it much easier to browse these results. In the past, clustering-based methods have been applied to solve the search-result organization problem, but it has been difficult to extract the human-readable descriptions for these clusters. An alternative solution to this problem is to generate a series of labels from search results firstly, and then assign documents to relevant labels to form labeled categories. In this approach, a major task is how to generate the labels for the documents. In this paper, we propose a novel label generation method: firstly, we extract some phrases as candidates of labels based on the search results, and adopt a binary classifier as our learning model to classify these label candidates into useful or meaningless label category. Then, the candidates in the useful label category form the final results. As our method is applied on the search results which are retrieved from a domain-specified corpus instead of general corpus, there're some special features of the labels for classification. Experimental results show that the accuracy of our system is nearly 10% higher than using the mutual information criterion, which is an unsupervised method for solving this problem, to do the label selection