Query Classification by Leveraging Explicit Concept Information

This paper presents a hybrid method for user query reformation and classification depending on fuzzy semantic-based approach and KNN classifier. The overall processes of the system are: query preprocessing, fuzzy membership calculation and query classification and reformation. Classification is performed using KNN classifier not just by keyword-based semantic but using a sentence-level semantics. A semantic model - EnrichSearch - has been developed for applying this approach. After classification, user's query is reformulated to be submitted to a search engine which gives better results than submitting the original query to the search engine. Experiments show significant enhancement in search results over traditional keyword-based search engines' results.

Query understanding is a fundamental problem in information retrieval (IR), which has attracted continuous attention through the past decades. Many different tasks have been proposed for understanding users' search queries, e.g., query classification or query clustering. However, it is not that precise to understand a search query at the intent class/cluster level due to the loss of many detailed information. As we may find in many benchmark datasets, e.g., TREC and SemEval, queries are often associated with a detailed description provided by human annotators which clearly describes its intent to help evaluate the relevance of the documents. If a system could automatically generate a detailed and precise intent description for a search query, like human annotators, that would indicate much better query understanding has been achieved. In this paper, therefore, we propose a novel Query-to-Intent-Description (Q2ID) task for query understanding. Unlike those existing ranking tasks which leverage the query and its description to compute the relevance of documents, Q2ID is a reverse task which aims to generate a natural language intent description based on both relevant and irrelevant documents of a given query. To address this new task, we propose a novel Contrastive Generation model, namely CtrsGen for short, to generate the intent description by contrasting the relevant documents with the irrelevant documents given a query. We demonstrate the effectiveness of our model by comparing with several state-of-the-art generation models on the Q2ID task. We discuss the potential usage of such Q2ID technique through an example application.

The task of topical classification of Web queries is to classify Web queries into a set of target categories. Machine learning based conventional approaches usually rely on external sources of information to obtain additional features for Web queries and training data for target categories. Unfortunately, these approaches are known to suffer from inability to adapt to different target categories which may be caused by the dynamic changes observed in both Web topic taxonomy and Web content. In this paper, we propose a feature-free flexible approach to topical classification of Web queries. Our approach analyzes queries and topical categories themselves and utilizes the number of Web pages containing both a query and a category to determine their similarity. The most attractive feature of our approach is that it only utilizes the Web page counts estimated by a search engine to provide the Web query classification with respectable accuracy. We conduct experimental study on the effectiveness of our approach using a set of rank measures and show that our approach performs competitively to some popular state-of-the-art solutions which, however, make frequent use of external sources and are inherently insufficient in flexibility.

Queries describe the users' search intent and therefore they play an essential role in the context of ranking for information retrieval and Web search. However, most of existing approaches for ranking do not explicitly take into consideration the fact that queries vary significantly along several dimensions and entail different treatments regarding the ranking models. In this paper, we propose to incorporate query difference into ranking by introducing query-dependent loss functions. In the context of Web search, query difference is usually represented as different query categories; and, queries are usually classified according to search intent such as navigational, informational and transactional queries. Based on the observation that such kind of query categorization has high correlation with the user's different expectation on the result accuracy on different rank positions, we develop position-sensitive query-dependent loss functions exploring such kind of query categorization. Beyond the simple learning method that builds ranking functions with pre-defined query categorization, we further propose a new method that learns both ranking functions and query categorization simultaneously. We apply the query-dependent loss functions to two particular ranking algorithms, RankNet and ListMLE. Experimental results demonstrate that query-dependent loss functions can be exploited to significantly improve the accuracy of learned ranking functions. We also show that the ranking function jointly learned with query categorization can achieve better performance than that learned with pre-defined query categorization. Finally, we provide analysis and conduct additional experiments to gain deeper understanding on the advantages of ranking with query-dependent loss functions over other query-dependent ranking approaches and query-independent approaches.

Search engines usually return relevant sorted results based on the keywords. Because of the lack of considering the user's current search interest and intention, this kind of strategy may not meet users' personalized search requirements. In order to retrieve results associated with the user's current search interest, existing researches integrate the user's interest information into the search process and ranking, so as to make the search results fit to the user's current search target more better. Different from the unified model of interest which most current related researches used, we propose a model of combing user's interests for the purpose of ranking in the light of different query types. This paper first classified queries into several types based on the current keywords and the sources of user's interests, such as short-term interest in user clicks, and long-term interest in user's search history, search logs of the search system etc. Then based on the sources of interests, we established a model of combination of interests which based on the artificial neural network and was trained by related approaches and data sets. Depending on different query types, we combined these sources of user's interests that the most best reflecting the user's current search intention to enhance the user's search experience. Comparing with the existing model of single interest, the results of the experiments show that our model can obtain better ranking on the results.

As the amount of information is growing rapidly on world wide web, it has become very difficult to get relevant information using traditional search engines within a stipulated time. The main reasons for irrelevant search results are the lack of understanding of user's search intention or user's preferences, keyword based searching, short queries. In this paper, we will study different features that are used in information retrieval. We will also discuss various machine learning techniques that are helpful in deciding the relevance of web page to user. We have done classification on the basis of features. In the end we will compare different techniques and their pros and cons are also discussed.

Click-through data has proven to be a valuable resource for improving search-ranking quality. Search engines can easily collect click data, but biases introduced in the data can make it difficult to use the data effectively. In order to measure the effects of biases, many click models have been proposed in the literature. However, none of the models can explain the observation that users with different search intent (e.g., informational, navigational, etc.) have different click behaviors. In this paper, we study how differences in user search intent can influence click activities and determined that there exists a bias between user search intent and the relevance of the document relevance. Based on this observation, we propose a search intent bias hypothesis that can be applied to most existing click models to improve their ability to learn unbiased relevance. Experimental results demonstrate that after adopting the search intent hypothesis, click models can better interpret user clicks and substantially improve retrieval performance.

Learning to understand user search intents from their online behaviors is crucial for both Web search and online advertising. However, it is a challenging task to collect and label a sufficient amount of high quality training data for various user intents such as compare products, plan a travel, etc. Motivated by this bottleneck, we start with some user common sense, i.e. a set of rules, to generate training data for learning to predict user intents. The rule-generated training data are however hard to be used since these data are generally imperfect due to the serious data bias and possible data noises. In this paper, we introduce a Co-learning Framework (CLF) to tackle the problem of learning from biased and noisy rule-generated training data. CLF firstly generates multiple sets of possibly biased and noisy training data using different rules, and then trains the individual user search intent classifiers over different training datasets independently. The intermediate classifiers are then used to categorize the training data themselves as well as the unlabeled data. The confidently classified data by one classifier are added to other training datasets and the incorrectly classified ones are instead filtered out from the training datasets. The algorithmic performance of this iterative learning procedure is theoretically guaranteed.

In the last few years, we have witnessed an explosive growth in multimedia content. Online repositories such as Flickr (image) and YouTube (video) contain hundreds of millions of images and videos and are still growing by the day. However, the utility of these data sources is only as good as their accessibility. A primary motivation of this thesis is towards better utilization of this overwhelming wealth of information. Specifically, we look at ways to disambiguate search in a multimedia retrieval system. The need for search disambiguation often arises in practice because the typical issued query is short, imprecise and ambiguous, and it is difficult to infer the exact query intent. Search disambiguation is the general class of techniques to elucidate the multi-faceted information needs arising from an ambiguous query. The objective is to provide an overview of search results grouped by multiple facets, which can better clarify a user's search intention by enabling him to zoom into any specific facet of the query topic. There has been a lack of a principled approach to search disambiguation. Many multimedia retrieval systems address the issue of ambiguous queries by results clustering, near-duplicate removal and diversification. The aim is to simply prevent result pages from being cluttered by too many similar Web articles. In this thesis, we take a more principled approach and propose the following two-prong methodology. First, we propose a novel faceted topic retrieval framework wherein multimedia documents are modeled as comprising of facets or topics. The notion of facets can be interpreted broadly to encompass any binary property of a document that represents a fact or a topic that is contained in the query need. In contrast with traditional term-based retrieval models (e.g. TF-IDF), documents are now ranked by the relevance of their composite facets/topics to the query. The goal is then to return a set of documents that are not only relevant to the query, they also cover the many different facets of the information need. We base our faceted retrieval framework on probabilistic topic models, a class of algorithms designed to discover the latent thematic structures in a document collection. Second, we augment our faceted retrieval framework with two modeling capabilities to better process ambiguous queries in multimedia collection such as video and images. Firstly, because different queries may have varying level of ambiguity, and hence varying level of polysemy in the return results, we develop a non-parametric Bayesian method to cluster search results. The main advantage of our clustering method is that the number of mixture components is not fixed a priori, but is determined during the posterior inference process. This allows our model to grow with the level of polysemy (and visual diversity) in the return multimedia results. Secondly, we extend the basic probabilistic topic model (the Latent Dirichlet Allocation, LDA) to jointly model the complementary information in the visual and textual streams. We show how the joint modeling can improve the quality of facet detection in news video, and in turn yield better user satisfaction in faceted topic retrieval.

Information Retrieval Systems usually employ syntactic search techniques to match a set of keywords with the indexed content to retrieve results. But pure keyword-based matching lacks on capturing user's search intention and context and suffers of natural language ambiguity and vocabulary mismatch. Considering this scenario, the hypothesis raised is that the use of embeddings in a semantic search approach will make search results more meaningfully. Embeddings allow to minimize problems arising from terminology and context mismatch. This work proposes a semantic similarity function to support semantic search based on hyper relational knowledge graphs. This function uses embeddings in order to find the most similar nodes that satisfy a user query.

Keyword search on XML data has been a hot research issue recently. Towards the ultimate goal of retrieving results that match user's search intention, existing methods keep improving the matching semantics and result retrieval methods. A list of query results in the form of subtrees will be returned to the users. However, we find that the traditional way of returning a list of subtrees to the users is still not sufficient to meet users' information needs because: (1) search intention of a keyword query can be different with different users issuing it; (2) amongst the query results, they may have sibling and containment relationships, which could be important for digesting the query results and should be displayed to the users as well. To address the problem, we propose a new exploration model MALEX to work as a complementary component of the XML keyword search engine, in order to enhance users' search experience. It can even work independently as a new way to explore the XML database.

Poor retrieval performance significantly degenerates users' experience of visual search, especially in mobile search. Ideally, users would like to be alerted when bad queries are present, which helps eliminate latency as well as waste of bandwidth, especially in 3G wireless environment. In this paper, we propose a visual query performance prediction (v-QPP) approach to predict the retrieval effectiveness. We employ latent dirichlet allocation (LDA)to derive latent topics from image database. From the collection statistics, we model the query's specificity based on topics. High specificity helps a retrieval system to derive user's search intent exactly. Moreover, as low discriminative content is difficult to search in terms of distinguishing relevant images from irrelevant one, we propose a topics based inverse concept frequency (t-ICF) model to deal with specific queries but difficult to discriminate in the reference database. Comparison experiments over MPEG CDVS benchmarking datasets have shown our method significantly outperforms existing approaches in document retrieval.

Text document classification is an important re-search topic in the field of information retrieval, and so it is how we represent the information extracted from the documents to be classified. There exists document classification methods and tech-niques based on the vector space model, which doesn’t capture the relation between words, which is considered of importance to make a better comparison and therefore classification. For this reason, two significant contributions were made, the first one is the way to create the feature vector for document comparison, which uses adapted concepts of non-rigid 3D models comparison and graphs as a data structure to represent such documents. The second contribution is the classification method itself, which uses the representative feature vectors of each category to classify new documents.

Understanding Web users' search intent expressed by their queries is essential for a search engine to provide the appropriate answers. Web query classification (QC) algorithms have been widely studied to improve the accuracy and meet users' demands. Some QC algorithms convert queries into vectors and use SVM or CRF model as the classifier. However, with the volume of data increasing, the time consumed significantly increases. In this paper, we propose a method in which we split the queries into words and convert queries into a graph, after that, we adopt a liner equation as the classifier. Experimental results exhibit that our method has similar accuracy but higher efficiency compared with the existing methods. Our method can decrease the training time by 10% compared with the SVM algorithm, and also outperform the CRF model.

Developer communities built around software products, like the SAP Community Network, provide a knowledge base for reocurring problems and their solutions. Due to the large amount of content maintained in such communities, e.g., in forums, finding relevant solutions is a major challenge beyond the scope of common keyword-based search engines. In fact, it is measured that around 50% of the forum questions of our particular scenario have already been answered at the time they are posted. We target this challenge by an entity aware search, which exploits structured knowledge, such as domain-specific ontologies, for both query interpretation and creation of document indexes. The system takes a natural language query as input, interprets it as an entity graph, matches this graph with pre-processed content and supports the user in refining his query based on the top-k relevant entities. Results are presented in a user interface that supports faceted search based on entities. Additionally, the user interface is structured according to possible search intentions of users. The evaluation of our system on the SCN scenario yields that the top 5 entities in user queries are recognized with a precision of 83% compared to 61% of state of the art algorithms.