We propose a tool for experts finding applied to academic data generated by the start-up DSRT in the context of its application Peerus. A user may submit the title, the abstract and optionnally the authors and the journal of publication of a scientific article and the application then returns a list of experts, potential reviewers of the submitted article. The retrieval algorithm is a voting system based on a language modeling technique trained on several millions of scientific papers.

Expertise is a loosely defined concept that is hard to formalize. Much research has focused on designing efficient algorithms for expert finding in large databases in various application domains. The evaluation of such recommender systems lies most of the time on human-annotated sets of experts associated with topics. The protocol of evaluation consists in using the namings or short descriptions of these topics as raw queries in order to rank the available set of candidates. Several measures taken from the field of information retrieval are then applied to rate the rankings of candidates against the ground truth set of experts. In this paper, we apply this topic-query evaluation methodology with the AMiner data and explore a new document-query methodology to evaluate experts retrieval from a set of queries sampled directly from the experts documents. Specifically, we describe two datasets extracted from AMiner, three baseline algorithms from the literature based on several document representations and provide experiment results to show that using a wide range of more realistic queries provides different evaluation results to the usual topic-queries.

Nous proposons un outil de recherche d’experts applique au monde academique sur les donnees generees par l’entreprise DSRT dans le cadre de son application Peerus 1 . Un utilisateur soumet le titre, le resume et optionnellement les auteurs et le journal de publication d’un article scientifique et se voit proposer une liste d’experts, potentiels reviewers de l’article soumis. L’algorithme de recherche est un systeme de votes reposant sur un modele du langage entraine a partir d’un ensemble de plusieurs millions d’articles scientifiques. L’outil est accessible a chacun sous la forme d’une application web intitulee Peerus Review