Extraction d'entit\'es dans des collections \'evolutives

TL;DR

This paper presents a method for extracting named entities, specifically partner names, from evolving collections of reports using syntactic patterns and supervised learning, without relying on linguistic resources.

Contribution

It introduces an approach that combines pattern discovery and supervised validation for entity extraction in dynamic document collections, avoiding the need for extensive training data.

Findings

Extraction performance improves with larger training sets

Method does not require linguistic resources or large annotated datasets

Approach adapts to evolving document collections

Abstract

The goal of our work is to use a set of reports and extract named entities, in our case the names of Industrial or Academic partners. Starting with an initial list of entities, we use a first set of documents to identify syntactic patterns that are then validated in a supervised learning phase on a set of annotated documents. The complete collection is then explored. This approach is similar to the ones used in data extraction from semi-structured documents (wrappers) and do not need any linguistic resources neither a large set for training. As our collection of documents would evolve over years, we hope that the performance of the extraction would improve with the increased size of the training set.