CYUT 團隊參加社群書籍搜尋國際公開評測系統從INEX2013 到CLEF2016

The CYUT System on Social Book Search Track since INEX 2013 to CLEF 2016

The Social Book Search (SBS) Lab is part of Conference and Labs of the Evaluation Forum(CLEF) lab series, which provides query topics on book suggestion in natural language.Participant teams have to build systems that can make suggestions out of 2.8 million books.This paper reports how the CYUT team attends the SBS track from 2013 to 2016. Our system isbased on keyword searching and ranking by social features. We also design a query expansionmodule which is based on word2vec, a word embedding toolkit. The new module helps oursystem to get better performance in suggestion track. This paper reports the progression of oursystem.

本論文報告朝陽科技大學參加INEX 2013, 2014 以及後續歐洲資訊檢索論壇CLEF 2015～2016 舉辦的SBS 開發經驗。這個評測的目標是由系統從280 萬本書中自動推薦相關圖書給使用者。主辦單位蒐集了許多使用者在社群媒體上對書籍推薦的需求，形成一個測試集。所有參加公開評測的系統回傳推薦圖書之後，由主辦單位確認是否是恰當的推薦，藉此評估系統的優劣。我們的系統整合了搜尋、排序、語意理解等技術，逐年改良了系統的效能。