Short Paper: Improving the archiving and contextualization of electronic messaging in French

Abstract

The objective of the Pêle-mél program is to propose a prototype tool for exploring and visualizing electronic messages and to test different strategies for contextualizing and classifying electronic messages in French, using NLP techniques, artificial intelligence and learning. Beyond that, the aim is also to draw conclusions on archiving strategies and to theorize the scope of the external knowledge required to succeed in this type of project. If the relevance of NLP techniques is proven to exploit the information content of e-mails, if experiments have already been conducted in English, this program, financed by the French Ministry of Culture, innovates on two points: the will to understand messaging systems in their reticular context and the deployment of techniques adapted to French. The first results confirm, thanks to a detailed analysis of the messages, the interest of large-scale archiving. They also show the difficulties linked to the temporal context of the studied mailboxes and to the hybrid character of the information supports. The palliative strategies put in place are costly but possible.

Details

Creators

Edouard Vasseur

Institutions

Ecole Nationale Des Chartes

Date

Keywords

email; machine learning; classification; terminology

Publication Type

short paper

License

CC-BY 4.0 International

Download

(unknown) bytes

View This Publication