Resultados de pesquisa

This QA4MRE edition brought two challenges to the DI@UE team: the absence of Portuguese as a working language and the different nature of the task when compared with previous participation in QA@CLEF. We addressed this multiple choice answering problem by assessing answer candidates in a text surface based manner, without a deep linguistic processing. This system employs a Lucene based search engine and Wordnet to assist in synonym check and morphological normalization. Answer analysis and the criteria for the answering decision are fun- damentally based on superficial analysis of document text, enriched with semantic validation of compatibility between terms. The solution we describe answered to 73 from 120 questions, having 18 correct answers and an accuracy of 0.15.

In the 2012 edition of CLEF, the DI@UE team has signed up for Question Answering for Machine Reading Evaluation (QA4MRE) main task. For each question, our system tries to guess which of the five hypotheses is the more plausible response, taking into account the reading test content and the documents from the background collection on the question topic. For each question, the system applies Named Entity Recognition, Question Classification, Document and Passage Retrieval. The criteria used in the first run is to choose the answer with the smallest distance between question and answer key elements. The system applies a specific treatment for certain factual questions, with the categories Quantity, When, Where, What, and Who, whose responses are usually short and likely to be detected in the text. For the second run, the system tries ...

This article describes the participation of a group from the University of Évora in the CLEF2013 QA4MRE main task. Our system has a superficial text analysis based approach. The methodology starts with the preprocessing of background collection documents, whose texts are lemmatized and then indexed. Named entities and numerical expressions are sought in questions and their candidate answers. Then the lemmatizer is applied and stop words are removed. Answer patterns are formed for each question+answer pair, with a search query for document retrieval. Original search terms are expanded with synonyms and hyperonyms. Finally, the texts retrieved for each candidate response are segmented and scored for answer selection. Considering only the main questions, the system best result was obtained in the third run, having answered to 206 quest...

The diue system uses a supervised Machine Learning approach for the polarity classification subtask of RepLab. We used the Python NLTK for preprocessing, including file parsing, text analysis and feature extraction. Our best solution is a mixed strategy, combining bag-of-words with a limited set of features based on sentiment lexicons and superficial text analysis. This system begins by applying tokenization and lemmatization. Then each tweet content is analyzed and 18 features are obtained, related to presence of polarized term, negation before polarized expression and entity reference. For the first run, the learning and classification were performed with the Decision Tree algorithm, from the NLTK framework. In the second run, we used a pipeline of classifiers. The first classifier applies Naive Bayes in a bag-of-words feature mod...