White Papers & Technical Reports
Diversity In Expert Search
Diversity in Expert Search. When Web search engines deal with a broad or ambiguous search query from a user, it is often better to try to present in the results documents matching various aspects of the query to increase the user satisfaction. Is this practice specific to Web search, or can it be applied to expert search as well?
When a user has a technological or scientific need and searches for experts, there are several aspects in which diversification of results can be useful. First, the need may be ambiguous or related to more than one scientific discipline. Hence, the expert search engine must ensure that the related disciplines are represented in the results. Second, a user may be searching for experts in a broad geographic region. Consequently, geographic diversity can be achieved by matching experts from several organizations and institutes. Third, a user may be looking for experts that are more likely to collaborate with the industry. Such experts are more likely to have patents, so the expert search engine must weigh accordingly such evidence in addition to mentions on academic Web sites or scientific publications when ranking experts.
Diversifying results in Expert Search. Ranking experts for a query is typically a three-stage process. In the first stage, documents matching the query are retrieved and ranked. In the second stage, experts are associated to the ranked documents. In the third stage, a score is computed for each expert based on the associated documents, and a ranked list of experts is produced.
Methods to promote diversity of results can be applied at each of the three stages with different implications about the final ranking. When diversifying the initial document ranking, the final ranking favors people with cross-field expertise. When directly diversifying the matched experts, then the final ranking will favor experts specialized in distinct subtopics. Finally, diversifying the supporting documents of each ranked expertise expected to boost in the final ranking experts with a variety of supporting documents. For example, an expert who has both published articles and filed patents on a topic, or someone who has been active in different countries would obtain more associated documents.
A Novel Approach for Open Innovation Platforms:Multistep Dynamic Expert Sourcing
Open Innovation platforms (and especially problem solving platforms) have received increasing attention over the last few years. Although much hope was initially put into them, it gradually faded away and changed in some form of mistrust, because of a number of drawbacks (relatively low number of solvers, confidentiality issues, intellectual property management problems, disappointing solutions, etc.). We propose a novel approach, Multistep Dynamic Expert Sourcing (MDES), developed and implemented by the French company PRESANS. This approach solves most of the existing drawbacks and leverages all the potential of online Open Innovation.