Abstract
User's profiles play an important role when information systems try to meet their needs. This work presents a novel approach to build user profiles. It is based on information extraction techniques and proceeds by iterative steps. The use of different statistic metrics, Natural Language Processing (NLP) techniques and semantic descriptions (ontologies) in the authors' approach, has provided it with a good precision degree when extracting information from texts. This has been demonstrated by an application prototype which is an automatic user profile constructor, using the texts of emails job applications (E recruitment field).Article Preview
TopInformation Extraction (IE) aims at exploring and exploiting various formats of data. It uses a set of techniques and methods to extract relevant information that can be conveyed by an information source. Several research studies have emerged, in this section we present a classification of the most relevant works according to their application field:
2.1. Automatic Summarization
An automatic summarization system aims at returning a condensed representation of an original text while keeping its semantic. In other words, it aims at extracting relevant information. Several summarization techniques have been developed and allow the construction of several summarization systems such (Torres & Rodrigez, 2010; Bossard, 2010; Boudin, 2008; Erkan & Radev, 2004), YACHS2, CORTEX and ROUGE2.
They use a set graph theory techniques and statistical metrics. Once assembled, they allow to assign a score to each sentence of the document, then, sentences having the highest scores are selected to be a part of the summary.