SEUPD@CLEF: Team Gamora on Argument Retrieval for Controversial Questions

This paper is the report of the work we have done for Argument Retrieval CLEF 2022 Touché Task 1 as Gamora team. Argument Retrieval CLEF 2022 Touché Task 1 focuses on the problem of retrieving and ranking relevant pairs of sentences from a collection of arguments for a given controversial questions. After an analysis of the structure and the possibilities of manipulating document data, we concentrated our work on query management, applying methods of query expansion, reduction, query boost to conclude with satisfactory results with the last two solutions listed. We came up with two systems, one based on a two-stage operation, double index and double search, and one based on sentence pair ranking based on its argumentative quality, assessed by a machine learning model.