Taxonomy-Driven Expertise Aggregation on Bibliographic Data

With the enormous growth of data, retrieving information from the Web became more desirable and even more challenging because of the Big Data issues (e.g. noise, corruption, bad quality…etc.). Expert seeking, defined as returning a ranked list of expert researchers given a topic, has been a real concern in the last 15 years. This kind of task comes in handy when building scientific committees, requiring to identify the scholars' experience to assign them the most suitable roles in addition to other factors as well. Due to the fact the Web is drowning with plenty of data, this opens up the opportunity to collect different kinds of expertise evidence. In this paper, we propose an expert seeking approach with specifying the most desirable features (i.e. criteria on which researcher's evaluation is done) along with their estimation techniques. We utilized some machine learning techniques in our system and we aim at verifying the effectiveness of incorporating influential features that go beyond publications.

2017

In this work, we discuss enhanced techniques that optimize expert representation and identify subject experts via clustering analysis of the available online information. We use a weighting method to assess the levels of expertise of an expert to the domain-specific topics. In this context, we define a way to estimate the expertise similarity between experts. Then the experts finding task is viewed as a list completion task and techniques that return similar experts to ones provided by the user are considered. In addition, we discuss a formal concept analysis approach for clustering of a group of experts with respect to given subject areas. The produced grouping of experts can further be used to identify individuals with the required competence.

Proceedings of the 25th International Conference on World Wide Web, 2016

We introduce an unsupervised discriminative model for the task of retrieving experts in online document collections. We exclusively employ textual evidence and avoid explicit feature engineering by learning distributed word representations in an unsupervised way. We compare our model to state-of-the-art unsupervised statistical vector space and probabilistic generative approaches. Our proposed log-linear model achieves the retrieval performance levels of state-of-the-art document-centric methods with the low inference cost of so-called profile-centric approaches. It yields a statistically significant improved ranking over vector space and generative models in most cases, matching the performance of supervised methods on various benchmarks. That is, by using solely text we can do as well as methods that work with external evidence and/or relevance feedback. A contrastive analysis of rankings produced by discriminative and generative approaches shows that they have complementary strengths due to the ability of the unsupervised discriminative model to perform semantic matching.

22nd International Conference on Electronic Publishing

ArXiv, 2015

Finding experts for a given problem is recognized as a difficult task. Even when a taxonomy of subject expertise exists, and is associated with a group of experts, it can be hard to exploit by users who have not internalized the taxonomy. Here we present a method for both attaching experts to a domain ontology, and hiding this fact from the end user looking for an expert. By linking Wikipedia to this same pivot ontology, we describe how a user can browse Wikipedia, as they normally do to search for information, and use this browsing behavior to find experts. Experts are characterized by their textual productions (webpages, publications, reports), and these textual productions are attached to concepts in the pivot ontology. When the user finds the Wikipedia page characterizing their need, a list of experts is displayed. In this way we transform Wikipedia into a search engine for experts.

2019

Today there is no unified model for selecting experts or specialists on formal grounds, therefore this project proposes an approach to solving the problem of searching for experts based on networks of co-authors. The method of research of scientific collectives is proposed, which allows to find groups of scientists whose activities correspond to the topics on the basis of the network of terms according to the given terms of a certain domain. The work of the algorithm is realized on the basis of information from the Web of science, a comparison of the work of the algorithm with the system Aminer. The proposed method allows searching for more flexible requests. It is advisable to use the authors' databases while searching for scientific groups to exclude identical names and obtain detailed information.

The Semantic Web: Research and Applications, 2010

2011

Abstract. The task of expert finding has been getting increasing attention in information retrieval literature. However, the current stateof-the-art is still lacking in principled approaches for combining different sources of evidence.

2007

We provide a benchmark dataset for expert finding within the computer science domain. We show how large isolated data graphs from disparate structured data sources can be combined to form one, large, well-linked RDF graph and implement these methods to achieve our dataset. Such a graph lends itself to links analysis and thus opens up possibilities for analysis by expert finding techniques.

2007

This paper shows a method for finding topic-centric experts from open access metadata and full text documents. Topic-centric information including experts is served on OntoFrame, which is a Semantic Web-based academic research information service supporting R&D activities. URI schemebased OntoFrame provides three entity pages: topic, person, and event. 'Persons by Topic' in topic page lists up topic-centric identified experts. SPARQL query is used to retrieve them from RDF triple store through backward chaining.