Showing papers on "Question answering published in 2005"

Context-sensitive information retrieval using implicit feedback

Abstract: A major limitation of most existing retrieval models and systems is that the retrieval decision is made based solely on the query and document collection; information about the actual user and search context is largely ignored. In this paper, we study how to exploit implicit feedback information, including previous queries and clickthrough information, to improve retrieval accuracy in an interactive information retrieval setting. We propose several context-sensitive retrieval algorithms based on statistical language models to combine the preceding queries and clicked document summaries with the current query for better ranking of documents. We use the TREC AP data to create a test collection with search context information, and quantitatively evaluate our models using this test set. Experiment results show that using implicit feedback, especially the clicked document summaries, can improve retrieval performance substantially.

Support Vector Learning for Semantic Argument Classification

Abstract: The natural language processing community has recently experienced a growth of interest in domain independent shallow semantic parsing--the process of assigning a Who did What to Whom, When, Where, Why, How etc. structure to plain text. This process entails identifying groups of words in a sentence that represent these semantic arguments and assigning specific labels to them. It could play a key role in NLP tasks like Information Extraction, Question Answering and Summarization. We propose a machine learning algorithm for semantic role parsing, extending the work of Gildea and Jurafsky (2002), Surdeanu et al. (2003) and others. Our algorithm is based on Support Vector Machines which we show give large improvement in performance over earlier classifiers. We show performance improvements through a number of new features designed to improve generalization to unseen data, such as automatic clustering of verbs. We also report on various analytic studies examining which features are most important, comparing our classifier to other machine learning algorithms in the literature, and testing its generalization to new test set from different genre. On the task of assigning semantic labels to the PropBank (Kingsbury, Palmer, & Marcus, 2002) corpus, our final system has a precision of 84% and a recall of 75%, which are the best results currently reported for this task. Finally, we explore a completely different architecture which does not requires a deep syntactic parse. We reformulate the task as a combined chunking and classification problem, thus allowing our algorithm to be applied to new languages or genres of text for which statistical syntactic parsers may not be available.

Question answering passage retrieval using dependency relations

Abstract: State-of-the-art question answering (QA) systems employ term-density ranking to retrieve answer passages Such methods often retrieve incorrect passages as relationships among question terms are not considered Previous studies attempted to address this problem by matching dependency relations between questions and answers They used strict matching, which fails when semantically equivalent relationships are phrased differently We propose fuzzy relation matching based on statistical models We present two methods for learning relation mapping scores from past QA pairs: one based on mutual information and the other on expectation maximization Experimental results show that our method significantly outperforms state-of-the-art density-based passage retrieval methods by up to 78% in mean reciprocal rank Relation matching also brings about a 50% improvement in a system enhanced by query expansion

Question answering system, data search method, and computer program

Abstract: A question answering system includes a question answering unit, an analysis unit, a tree structure generation unit, a feature extraction unit, an evaluation unit and a re-ranking unit. The question answering unit executes search processing based on an input question and acquires answer candidates to the question. The analysis unit executes syntactic analysis processing or syntactic and semantic analysis processing on a passage obtained as a result of the search processing. The tree structure generation unit generates a tree structure based on an analysis result. The feature extraction unit extracts a relation between a search word applied in the search processing and each acquired answer candidate as a feature corresponding to each answer candidate from the tree structure. The evaluation unit determines an evaluation value of each answer candidate based on the feature extracted. The re-ranking unit re-ranks the answer candidates based on the evaluation values.

AquaLog: an ontology-portable question answering system for the semantic web

Abstract: As semantic markup becomes ubiquitous, it will become important to be able to ask queries and obtain answers, using natural language (NL) expressions, rather than the keyword-based retrieval mechanisms used by the current search engines. AquaLog is a portable question-answering system which takes queries expressed in natural language and an ontology as input and returns answers drawn from the available semantic markup. We say that AquaLog is portable, because the configuration time required to customize the system for a particular ontology is negligible. AquaLog combines several powerful techniques in a novel way to make sense of NL queries and to map them to semantic markup. Moreover it also includes a learning component, which ensures that the performance of the system improves over time, in response to the particular community jargon used by the end users. In this paper we describe the current version of the system, in particular discussing its portability, its reasoning capabilities, and its learning mechanism.

Finding experts in community-based question-answering services

Abstract: Software – User profiles and alert services.

Multilingual information access for text, speech and images : 5th workshop of the cross-language evaluation forum, CLEF 2004, Bath, UK, September 15-17, 2004, revised selected papers

Abstract: What Happened in CLEF 2004?.- What Happened in CLEF 2004?.- I. Ad Hoc Text Retrieval Tracks.- CLEF 2004: Ad Hoc Track Overview and Results Analysis.- Selection and Merging Strategies for Multilingual Information Retrieval.- Using Surface-Syntactic Parser and Deviation from Randomness.- Cross-Language Retrieval Using HAIRCUT at CLEF 2004.- Experiments on Statistical Approaches to Compensate for Limited Linguistic Resources.- Application of Variable Length N-Gram Vectors to Monolingual and Bilingual Information Retrieval.- Integrating New Languages in a Multilingual Search System Based on a Deep Linguistic Analysis.- IR-n r2: Using Normalized Passages.- Using COTS Search Engines and Custom Query Strategies at CLEF.- Report on Thomson Legal and Regulatory Experiments at CLEF-2004.- Effective Translation, Tokenization and Combination for Cross-Lingual Retrieval.- Two-Stage Refinement of Transitive Query Translation with English Disambiguation for Cross-Language Information Retrieval: An Experiment at CLEF 2004.- Dictionary-Based Amharic - English Information Retrieval.- Dynamic Lexica for Query Translation.- SINAI at CLEF 2004: Using Machine Translation Resources with a Mixed 2-Step RSV Merging Algorithm.- Mono- and Crosslingual Retrieval Experiments at the University of Hildesheim.- University of Chicago at CLEF2004: Cross-Language Text and Spoken Document Retrieval.- UB at CLEF2004: Cross Language Information Retrieval Using Statistical Language Models.- MIRACLE's Hybrid Approach to Bilingual and Monolingual Information Retrieval.- Searching a Russian Document Collection Using English, Chinese and Japanese Queries.- Dublin City University at CLEF 2004: Experiments in Monolingual, Bilingual and Multilingual Retrieval.- Finnish, Portuguese and Russian Retrieval with Hummingbird SearchServerTM at CLEF 2004.- Data Fusion for Effective European Monolingual Information Retrieval.- The XLDB Group at CLEF 2004.- The University of Glasgow at CLEF 2004: French Monolingual Information Retrieval with Terrier.- II. Domain-Specific Document Retrieval.- The Domain-Specific Track in CLEF 2004: Overview of the Results and Remarks on the Assessment Process.- University of Hagen at CLEF 2004: Indexing and Translating Concepts for the GIRT Task.- IRIT at CLEF 2004: The English GIRT Task.- Ricoh at CLEF 2004.- GIRT and the Use of Subject Metadata for Retrieval.- III. Interactive Cross-Language Information Retrieval.- iCLEF 2004 Track Overview: Pilot Experiments in Interactive Cross-Language Question Answering.- Interactive Cross-Language Question Answering: Searching Passages Versus Searching Documents.- Improving Interaction with the User in Cross-Language Question Answering Through Relevant Domains and Syntactic Semantic Patterns.- Cooperation, Bookmarking, and Thesaurus in Interactive Bilingual Question Answering.- Summarization Design for Interactive Cross-Language Question Answering.- Interactive and Bilingual Question Answering Using Term Suggestion and Passage Retrieval.- IV. Multiple Language Question Answering.- Overview of the CLEF 2004 Multilingual Question Answering Track.- A Question Answering System for French.- Cross-Language French-English Question Answering Using the DLT System at CLEF 2004.- Experiments on Robust NL Question Interpretation and Multi-layered Document Annotation for a Cross-Language Question/Answering System.- Making Stone Soup: Evaluating a Recall-Oriented Multi-stream Question Answering System for Dutch.- The DIOGENE Question Answering System at CLEF-2004.- Cross-Lingual Question Answering Using Off-the-Shelf Machine Translation.- Bulgarian-English Question Answering: Adaptation of Language Resources.- Answering French Questions in English by Exploiting Results from Several Sources of Information.- Finnish as Source Language in Bilingual Question Answering.- miraQA: Experiments with Learning Answer Context Patterns from the Web.- Question Answering for Spanish Supported by Lexical Context Annotation.- Question Answering Using Sentence Parsing and Semantic Network Matching.- First Evaluation of Esfinge - A Question Answering System for Portuguese.- University of Evora in QA@CLEF-2004.- COLE Experiments at QA@CLEF 2004 Spanish Monolingual Track.- Does English Help Question Answering in Spanish?.- The TALP-QA System for Spanish at CLEF 2004: Structural and Hierarchical Relaxing of Semantic Constraints.- ILC-UniPI Italian QA.- Question Answering Pilot Task at CLEF 2004.- Evaluation of Complex Temporal Questions in CLEF-QA.- V. Cross-Language Retrieval in Image Collections.- The CLEF 2004 Cross-Language Image Retrieval Track.- Caption and Query Translation for Cross-Language Image Retrieval.- Pattern-Based Image Retrieval with Constraints and Preferences on ImageCLEF 2004.- How to Visually Retrieve Images from the St. Andrews Collection Using GIFT.- UNED at ImageCLEF 2004: Detecting Named Entities and Noun Phrases for Automatic Query Expansion and Structuring.- Dublin City University at CLEF 2004: Experiments with the ImageCLEF St. Andrew's Collection.- From Text to Image: Generating Visual Query for Image Retrieval.- Toward Cross-Language and Cross-Media Image Retrieval.- FIRE - Flexible Image Retrieval Engine: ImageCLEF 2004 Evaluation.- MIRACLE Approach to ImageCLEF 2004: Merging Textual and Content-Based Image Retrieval.- Cross-Media Feedback Strategies: Merging Text and Image Information to Improve Image Retrieval.- ImageCLEF 2004: Combining Image and Multi-lingual Search for Medical Image Retrieval.- Multi-modal Information Retrieval Using FINT.- Medical Image Retrieval Using Texture, Locality and Colour.- SMIRE: Similar Medical Image Retrieval Engine.- A Probabilistic Approach to Medical Image Retrieval.- UB at CLEF2004 Cross Language Medical Image Retrieval.- Content-Based Queries on the CasImage Database Within the IRMA Framework.- Comparison and Combination of Textual and Visual Features for Interactive Cross-Language Image Retrieval.- MSU at ImageCLEF: Cross Language and Interactive Image Retrieval.- VI. Cross-Language Spoken Document Retrieval.- CLEF 2004 Cross-Language Spoken Document Retrieval Track.- VII. Issues in CLIR and in Evaluation.- The Key to the First CLEF with Portuguese: Topics, Questions and Answers in CHAVE.- How Do Named Entities Contribute to Retrieval Effectiveness?.

Knowledge representation and the semantics of natural language

Abstract: Knowledge Representation with MultiNet.- Historical Roots.- Basic Concepts.- Semantic Characterization of Objects.- Semantic Characterization of Situations.- The Comparison of Entities.- The Spatio-temporal Characterization of Entities.- Modality and Negation.- Quantification and Pluralities.- The Role of Layer Information in Semantic Representations.- Relations Between Situations.- Lexicon and Knowledge Representation.- Question Answering and Inferences.- Software Tools for the Knowledge Engineer and Sample Applications.- Comparison Between MultiNet and Other Semantic Formalisms or Knowledge Representation Paradigms.- The Representational Means of MultiNet.- Overview and Representational Principles.- Means for Expressing Classification and Stratification.- Relational and Functional Means of Representation.

RelExt : a tool for relation extraction from text in ontology extension

Abstract: Domain ontologies very rarely model verbs as relations holding between concepts. However, the role of the verb as a central connecting element between concepts is undeniable. Verbs specify the interaction between the participants of some action or event by expressing relations between them. In parallel, it can be argued from an ontology engineering point of view that verbs express a relation between two classes that specify domain and range. The work described here is concerned with relation extraction for ontology extension along these lines. We describe a system (RelExt) that is capable of automatically identifying highly relevant triples (pairs of concepts connected by a relation) over concepts from an existing ontology. RelExt works by extracting relevant verbs and their grammatical arguments (i.e. terms) from a domain-specific text collection and computing corresponding relations through a combination of linguistic and statistical processing. The paper includes a detailed description of the system architecture and evaluation results on a constructed benchmark. RelExt has been developed in the context of the SmartWeb project, which aims at providing intelligent information services via mobile broadband devices on the FIFA World Cup that will be hosted in Germany in 2006. Such services include location based navigational information as well as question answering in the football domain.

Evita: A Robust Event Recognizer For QA Systems

Abstract: We present Evita, an application for recognizing events in natural language texts. Although developed as part of a suite of tools aimed at providing question answering systems with information about both temporal and intensional relations among events, it can be used independently as an event extraction tool. It is unique in that it is not limited to any pre-established list of relation types (events), nor is it restricted to a specific domain. Evita performs the identification and tagging of event expressions based on fairly simple strategies, informed by both linguistic-and statistically-based data. It achieves a performance ratio of 80.12% F-measure.

Temporal and Event Information In Natural Language Text

Abstract: In this paper, we discuss the role that temporal information plays in natural language text, specifically in the context of question answering systems. We define a descriptive framework with which we can examine the temporally sensitive aspects of natural language queries. We then investigate broadly what properties a general specification language would need, in order to mark up temporal and event information in text. We present a language, TimeML, which attempts to capture the richness of temporal and event related information in language, while demonstrating how it can play an important part in the development of more robust question answering systems.

Retrieving answers from frequently asked questions pages on the web

Abstract: We address the task of answering natural language questions by using the large number of Frequently Asked Questions (FAQ) pages available on the web. The task involves three steps: (1) fetching FAQ pages from the web; (2) automatic extraction of question/answer (Q/A) pairs from the collected pages; and (3) answering users' questions by retrieving appropriate Q/A pairs. We discuss our solutions for each of the three tasks, and give detailed evaluation results on a collected corpus of about 3.6Gb of text data (293K pages, 2.8M Q/A pairs), with real users' questions sampled from a web search engine log. Specifically, we propose simple but effective methods for Q/A extraction and investigate task-specific retrieval models for answering questions. Our best model finds answers for 36% of the test questions in the top 20 results. Our overall conclusion is that FAQ pages on the web provide an excellent resource for addressing real users' information needs in a highly focused manner.

Multi-Perspective Question Answering Using the OpQA Corpus

Abstract: We investigate techniques to support the answering of opinion-based questions. We first present the OpQA corpus of opinion questions and answers. Using the corpus, we compare and contrast the properties of fact and opinion questions and answers. Based on the disparate characteristics of opinion vs. fact answers, we argue that traditional fact-based QA approaches may have difficulty in an MPQA setting without modification. As an initial step towards the development of MPQA systems, we investigate the use of machine learning and rule-based subjectivity and opinion source filters and show that they can be used to guide MPQA systems.

Analysis of Statistical Question Classification for Fact-Based Questions

Abstract: Question classification systems play an important role in question answering systems and can be used in a wide range of other domains. The goal of question classification is to accurately assign labels to questions based on expected answer type. Most approaches in the past have relied on matching questions against hand-crafted rules. However, rules require laborious effort to create and often suffer from being too specific. Statistical question classification methods overcome these issues by employing machine learning techniques. We empirically show that a statistical approach is robust and achieves good performance on three diverse data sets with little or no hand tuning. Furthermore, we examine the role different syntactic and semantic features have on performance. We find that semantic features tend to increase performance more than purely syntactic features. Finally, we analyze common causes of misclassification error and provide insight into ways they may be overcome.

Overview of the TREC 2005 Question Answering Track

Abstract: The TREC 2005 Question Answering track contained three tasks: the main question answering task, the document ranking task, and the relationship task. The main task was the same as the single TREC 2004 QA task. In the main task, question series were used to define a set of targets. Eac h series was about a single target and contained factoid and list questions. The final question in the series was an “Ot her” question that asked for additional information about the target that was not covered by previous questions in the series. The document ranking task was to return a ranked list of documents for each question from a subset of the questions in the main task, where the documents were thought to contain an answer to the question. In the relationship tas k, systems were given TREC-like topic statements that ended with a question asking for evidence for a particular relationship. The goal of the TREC question answering (QA) track is to foster research on systems that return answers themselves, rather than documents containing answers, in response to a question. The track started in TREC-8 (1999), with the first several editions of the track focused on factoid questions. A factoid question is a fact-based, short answer question such as How many calories are there in a Big Mac? . The task in the TREC 2003 QA track was a combined task that contained list and definition questions in additio n to factoid questions [1]. A list question asks for differen t instances of a particular kind of information to be returned , such as List the names of chewing gums . Answering such questions requires a system to assemble an answer from information located in multiple documents. A definition question asks for interesting information about a particul ar person or thing such as Who is Vlad the Impaler?or What is a golden parachute?. Definition questions also require systems to locate inform ation in multiple documents, but in this case the information of interest is much less crisply de lineated. The TREC 2004 test set contained factoid and list questions grouped into different series, where each series had the target of a definition associated with it [2]. Each questi on in a series asked for some information about the target. In addition, the final question in each series was an explicit “Other” question, which was to be interpreted as “Tell me other interesting things about this target I don’t know enou gh to ask directly”. This last question is roughly equivalen t to the definition questions in the TREC 2003 task. Several concerns regarding the TREC 2005 QA track were raised during the TREC 2004 QA breakout session. Since the TREC 2004 task was rather different from previous years’ tasks, there was the desire to repeat the task largely unchanged. There was also the desire to build infrastructure that would allow a closer examination of the role document retrieval techniques play in supporting QA technology. As a result of this discussion, the main task for the 2005 QA track was decided to be essentially the same as the 2004 task in that the test set would consist of a set of questio n series where each series asks for information regarding a particular target. As in TREC 2004, the targets included people, organizations, and other entities (things); unlike TREC 2004 the target could also be an event. Events were added since the document set from which the answers are to be drawn are newswire articles. The runs were evaluated using the same methodology as in TREC 2004, except that the primary measure was the per-series score instead of the combined component score. The document ranking task was added to the TREC 2005 track to address the concern regarding document retrieval and QA. The task was to submit, for a subset of 50 of the questions in the main task, a ranked list of up to 1000 documents for each question. Groups whose primary emphasis was document retrieval rather than QA, were allowed to participate in the document ranking task without submitting actual answers for the main task. However, all TREC 2005 submissions to the main task were required to include a ranked list of documents for each question in the document

Using Wikipedia at the TREC QA Track

Abstract: We describe our participation in the TREC 2004 Question Answering track. We provide a detailed account of the ideas underlying our approach to the QA task, especially to the so-called ?other? questions. This year we made essential use of Wikipedia, the free online encyclopedia, both as a source of answers to factoid questions and as an importance model to help us identify material to be returned in response to ?other? questions.

Question answering over structured content on the web

Abstract: Structured content and associated metadata from the Web are leveraged to provide specific answer string responses to user questions. The structured content can also be indexed at crawl-time to facilitate searching of the content at search-time. Ranking techniques can also be employed to facilitate in providing an optimum answer string and/or a top K list of answer strings for a query. Ranking can be based on trainable algorithms that utilize feature vectors for candidate answer strings. In one instance, at crawl-time, structured content is indexed and automatically associated with metadata relating to the structured content and the source web page. At search-time, candidate indexed structured content is then utilized to extract an appropriate answer string in response to a user query.

Overview of the CLEF 2005 multilingual question answering track

Abstract: The general aim of the third CLEF Multilingual Question Answering Track was to set up a common and replicable evaluation framework to test both monolingual and cross-language Question Answering (QA) systems that process queries and documents in several European languages. Nine target languages and ten source languages were exploited to enact 8 monolingual and 73 cross-language tasks. Twenty-four groups participated in the exercise. Overall results showed a general increase in performance in comparison to last year. The best performing monolingual system irrespective of target language answered 64.5% of the questions correctly (in the monolingual Portuguese task), while the average of the best performances for each target language was 42.6%. The cross-language step instead entailed a considerable drop in performance. In addition to accuracy, the organisers also measured the relation between the correctness of an answer and a system’s stated confidence in it, showing that the best systems did not always provide the most reliable confidence score. We provide an overview of the 2005 QA track, detail the procedure followed to build the test sets and present a general analysis of the results.

Learning to extract information from semi-structured text using a discriminative context free grammar

Abstract: In recent work, conditional Markov chain models (CMM) have been used to extract information from semi-structured text (one example is the Conditional Random Field [10]). Applications range from finding the author and title in research papers to finding the phone number and street address in a web page. The CMM framework combines a priori knowledge encoded as features with a set of labeled training data to learn an efficient extraction process. We will show that similar problems can be solved more effectively by learning a discriminative context free grammar from training data. The grammar has several distinct advantages: long range, even global, constraints can be used to disambiguate entity labels; training data is used more efficiently; and a set of new more powerful features can be introduced. The grammar based approach also results in semantic information (encoded in the form of a parse tree) which could be used for IR applications like question answering. The specific problem we consider is of extracting personal contact, or address, information from unstructured sources such as documents and emails. While linear-chain CMMs perform reasonably well on this task, we show that a statistical parsing approach results in a 50% reduction in error rate. This system also has the advantage of being interactive, similar to the system described in [9]. In cases where there are multiple errors, a single user correction can be propagated to correct multiple errors automatically. Using a discriminatively trained grammar, 93.71% of all tokens are labeled correctly (compared to 88.43% for a CMM) and 72.87% of records have all tokens labeled correctly (compared to 45.29% for the CMM).

Variations on language modeling for information retrieval

Abstract: Search engine technology builds on theoretical and empirical research results in the area of information retrieval (IR). This dissertation makes a contribution to the field of language modeling (LM) for IR, which views both queries and documents as instances of a unigram language model and defines the matching function between a query and each document as the probability that the query terms are generated by the document language model. The work described is concerned with three research issues.

Identifying Opinion Holders for Question Answering in Opinion Texts

Abstract: Question answering in opinion texts has so far mostly concentrated on the identification of opinions and on analyzing the sentiment expressed in opinions. In this paper, we address another important part of Question Answering (QA) in opinion texts: finding opinion holders. Holder identification is a central part of full opinion identification and can be used independently to answer several opinion questions such as “Is China supporting Bush’s war on Iraq?” and “Do Iraqi people want U.S. troops in their soil?”. Our system automatically learns the syntactic features signaling opinion holders using a Maximum Entropy ranking algorithm trained on human annotated data. Using syntactic parsing features, our system achieved 64% accuracy on identifying the holder of opinions in the MPQA dataset.

Analysis of polarity information in medical text.

Abstract: Knowing the polarity of clinical outcomes is important in answering questions posed by clinicians in patient treatment. We treat analysis of this information as a classification problem. Natural language processing and machine learning techniques are applied to detect four possibilities in medical text: no outcome, positive outcome, negative outcome, and neutral outcome. A supervised learning method is used to perform the classification at the sentence level. Five feature sets are constructed: unigrams, bigrams, change phrases, negations, and categories. The performance of different combinations of feature sets is compared. The results show that generalization using the category information in the domain knowledge base Unified Medical Language System is effective in the task. The effect of context information is significant. Combining linguistic features and domain knowledge leads to the highest accuracy.

Acquiring and Using World Knowledge using a Restricted Subset of English

Abstract: Many AI applications require a base of world knowledge to support reasoning. However, construction of such inferencecapable knowledge bases, even if constrained in coverage, remains one of the major challenges of AI. Authoring knowledge in formal logic is too complex a task for many users, while knowledge authored in unconstrained natural language is generally too difficult for computers to understand. However, there is an intermediate position, which we are pursuing, namely authoring knowledge in a restricted subset of natural language. Our claim is that this approach hits a “sweet spot” between the former two extremes, being both usable by humans and understandable by machines. We have developed such a language (called CPL, Computer-Processable Language), an interpreter, and a reasoner, and have used them to encode approximately 1000 “commonsense” rules (a mixture of general and domain-specific). The knowledge base is being used experimentally for semantic retrieval of video clips based on their captions, also expressed in CPL. In this paper, we describe CPL, its interpretation, and its use for reasoning, and discuss the strengths and weaknesses of restricted natural language as a the basis for knowledge representation.

Employing two question answering systems in TREC-2005

Abstract: In 2005, the TREC QA track had two separate tasks: the main task and the relationship task. To participate in TREC 2005 we employed two different QA systems. PowerAnswer-2 was used in the main task, whereas PALANTIR was used for the relationship questions. For the main task, new this year is the use of events as targets in addition to the nominal concepts used last year. Event targets ranged from a nominal event such as “Preakness 1998” to a description of an event as in “Plane clips cable wires in Italian resort”. There were 17 event targets total. Unlike nominal targets, which most often act as the topic of the subsequent questions, events provide a context for the questions. Therefore, targets representing events had questions that asked about participants in the event, about characteristics of the vent and furthermore, had temporal constraints. Also many questions referred to answers of previous questions. To complicate matters, several answers could be candidate for the anaphors used in follow-up questions, but salience mattered. This introduced new complexities for the coreference resolution. Consider the following example:

Computer-implemented system and method for providing authoritative answers to a general information search

Abstract: A computer-implemented system and method provides authoritative answers, developed within a community-based question answering service to users of a general network information search. This community-based question answering service receives a question from a first user, and receives answers from community members regarding this question. The authority of the answer is then determined by members of the community and if the authority is of an acceptable level, the question together with its authoritative answer is added to a database which includes all authoritatively answered questions. The answering service has an interface that exposes the contents of this database to queries from users of the network who are not necessarily members of the answering service. In one embodiment, results from queries of the community-based database are integrated with queries of a second database of general network information. An improved general information search service is also provided that includes query results from the authoritative answers generated by the community-based answering service.

Automated question answering: review of the main approaches

Abstract: Automated question-answering aims at delivering concise information that contains answers to user questions. This paper reviews and compares three main question-answering approaches based on natural language processing, information retrieval and question templates, eliciting their differences and the context of application that best suits each of them.

Statistical language modeling for information retrieval

Abstract: : This chapter reviews research and applications in statistical language modeling for information retrieval (IR) that has emerged within the past several years as a new probabilistic framework for describing information retrieval processes. Generally speaking, statistical language modeling, or more simply, language modeling (LM), refers to the task of estimating a probability distribution that captures statistical regularities of natural language use. Applied to information retrieval, language modeling refers to the problem of estimating the likelihood that a query and a document could have been generated by the same language model, given the language model of the document and with or without a language model of the query.

Generic soft pattern models for definitional question answering

Abstract: This paper explores probabilistic lexico-syntactic pattern matching, also known as soft pattern matching. While previous methods in soft pattern matching are ad hoc in computing the degree of match, we propose two formal matching models: one based on bigrams and the other on the Profile Hidden Markov Model (PHMM). Both models provide a theoretically sound method to model pattern matching as a probabilistic process that generates token sequences. We demonstrate the effectiveness of these models on definition sentence retrieval for definitional question answering. We show that both models significantly outperform state-of-the-art manually constructed patterns. A critical difference between the two models is that the PHMM technique handles language variations more effectively but requires more training data to converge. We believe that both models can be extended to other areas where lexico-syntactic pattern matching can be applied.

Answer determination for natural language questioning

Abstract: Open-domain question answering is the task of finding a concise answer to a natural language question using a large domain, such as the Internet. The use of a semantic role labeling approach to the extraction of the answers to an open domain factoid (Who/When/What/Where) natural language question that contains a predicate is described. Semantic role labeling identities predicates and semantic argument phrases in the natural language question and the candidate sentences. When searching for an answer to a natural language question, the missing argument in the question is matched using semantic parses of the candidate answers. Such a technique may improve the accuracy of a question answering system and may decrease the length of answers for enabling voice interface to a question answering system.

Experiments with Interactive Question-Answering

Abstract: This paper describes a novel framework for interactive question-answering (Q/A) based on predictive questioning. Generated off-line from topic representations of complex scenarios, predictive questions represent requests for information that capture the most salient (and diverse) aspects of a topic. We present experimental results from large user studies (featuring a fully-implemented interactive Q/A system named FERRET) that demonstrates that surprising performance is achieved by integrating predictive questions into the context of a Q/A dialogue.