Quantification et identification précises des concepts d'une requête Accurate and effective latent concept modeling for ad hoc information retrieval

International audienceDesigning approaches able to automatically detect uncertain expressions within natural language is central to design efficient models based on text analysis, in particular in domains such as question-answering, approximate reasoning, knowledge-based population. This article proposes an overview of several contributions and classifications defining the concept of uncertainty expressions in natural language, and the related detection methods that have been proposed so far. A new supervised and generic approach is next introduced for this specific task; it is based on the statistical analysis of multiple lexical and syntactic features used to characterize sentences through vector-based representations that can be analyzed by proven classification methods. The global performance of our approach is demonstrated and discussed with regard to various dimensions of uncertainty and text specificities.This method is available for download at https://github.com/pajean/uncertaintyDetection

Microblogging platforms such as Twitter are increasingly used for on-line client and market analysis. This motivated the proposal of a new track at CLEF INEX lab of Tweet Contextualization. The objective of this task was to help a user to understand a tweet by providing him with a short explanatory summary (500 words). This summary should be built automatically using resources like Wikipedia and generated by extracting relevant passages and aggregating them into a coherent summary. Running for four years, results show that the best systems combine NLP techniques with more traditional methods. More precisely the best performing systems combine passage retrieval, sentence segmentation and scoring, named entity recognition, text part-of-speech (POS) analysis, anaphora detection, diversity content measure as well as sentence reordering. This paper provides a full summary report on the four-year long task. While yearly overviews focused on system results, in this paper we provide a detailed report on the approaches proposed by the participants and which can be considered as the state of the art for this task. As an important result from the 4 years competition, we also describe the open access resources that have been built and collected. The evaluation measures for automatic summarization designed in DUC or MUC were not appropriate to evaluate tweet contextualization, we explain why and depict in detailed the LogSim measure used to evaluate informativeness of produced contexts or summaries. Finally, we also mention the lessons we learned and that it is worth considering when designing a task. (C) 2016 Elsevier Ltd. All rights reserved.

Working with corpus construction becomes an interesting alternative to different applications of natural language processing, such as, question-answering, machine translation, information retrieval, etc. Similarly, with the heterogeneous data and the user demands for the accurate information, many studies have accentuated the need of the Web to highlight the corpus construction. As well as, Arabic doesn’t have an equivalent number of linguistic corpuses as compared to other languages like English. In this paper, we focus on building our corpus of Arab questions-texts. We present a method for recovering text passages. This method is based on a real automatic interrogation of Google, in order to generate passages of texts and answer the factual questions. The first part of this paper describes the formal details about this method; the second part presents some experiments and results that validate our method