Back A people search engine based on equity and equality of opportunities

A people search engine based on equity and equality of opportunities

Thanks to the incorporation of an algorithm whose development has involved Carlos Castillo and Ricardo Baeza-Yates, coordinator and member, respectively, of the Web Science and Social Computing research group of the Department of Information and Communication Technologies, with help from the Data Transparency Lab, presented on 15 November at Telefonica Barcelona.

12.11.2018

 

People search engines are becoming increasingly common in personnel selection, to find self-employed workers or even to locate colleagues or friends. When searching for people, the top-k ranking algorithm is often used and the number of candidates that adapt to a search is usually high, and so often users do not pay attention to the entire list. In addition, conventionally, these lists are sorted in descending order based on some quality of the items (for example, years of experience or education, etc.).

The Conference will gather the best researchers, technologists, policy makers and industry representatives to pool advances in the ethics of artificial intelligence

On 15 November, in the framework of the 5th Conference of the Data Transparency Lab (DTL) at Telefonica Barcelona, Carlos Castillo, coordinator of the Web Science and Social Computing (WSSC) research group of the Department of Information and Communication Technologies (DTIC), will make a public presentation of a tool that implements the FA*IR algorithm that enhances justice and equality of opportunities in people search engines. The Conference will gather the best researchers, technologists, policy makers and industry representatives to pool advances in the ethics of artificial intelligence.

FA*IR was first presented at the Data Transparency Lab 2017 and the results of the research in the communication “FA*IR:  A Fair Top-k Ranking Algorithm” at the CIKM 2017 congress, held from 6 to 10 November in Singapore.

“Any developer can incorporate FA*IR into a search engine that is built with Elasticsearch”

“What I am going to show now at the 5th DTL Conference is how any developer can incorporate FA*IR into a search engine that is built with   Elasticsearch”, says Castillo. This has been possible thanks to UPF’s collaboration with Eurecat, ISI Foundation (Turin, Italy) and Technische Universität (TU, Berlin, Germany).

FA*IR is an electronic algorithm designed for search engines where the items being sought are people. With the help of 50,000 euros from the Data Transparency Lab 2017 it has been possible to implement FA*IR in one of today’s most used people search engines: Elasticsearch. 

“The algorithm incorporates a positive action policy to ensure that in all the results obtained according to the requested search there are people who belong to groups with some kind of disadvantage, minority or protected groups”

“In job searches, the recruiter may search according to different parameters: keywords, profile, etc., and get a list of people in accordance with the requested selection. The novelty of this algorithm is that it incorporates a positive action policy to ensure that, in all the results obtained according to the requested search, there are people who belong to groups with some kind of disadvantage, minority or protected group and also ensures that the presence of these people is not restricted only to the lowest positions of the ranking of candidates”, comments Carlos Castillo.

Multimedia

Categories:

SDG - Sustainable Development Goals:

Els ODS a la UPF

Contact