Improving Ecommerce Search with Query Named Entity Recognition
Nguyen, Dang (2020)
Nguyen, Dang
2020
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2020081219733
https://urn.fi/URN:NBN:fi:amk-2020081219733
Tiivistelmä
Site Search is an indispensable feature of any successful ecommerce businesses. Shopping experiences will be ruined if users cannot find what they are looking for. Ecommerce search expectation are high as driven by leading players like Amazon and Google. However, many search sites are falling to build a good search experience.
This thesis focuses on the query understanding component of the search – the key to unlock next level of search relevance. Query understanding is an active area of research, giving rises to different techniques that aim at understanding the search intent behind the search query.
Among many tasks of query understanding, Query Named Entity Recognition (QNER) aims to decode user intent by identifying and classifying query segments of the search queries. The QNER process is the enablement behind many query transformations tasks such as query scoping, query relaxation, query expansion. In addition, it will simplify the rest of informational retrieval process and open the opportunities for advanced features such as search suggestion, personalization, and recommendation.
The objective of this thesis is to build search system for ecommerce enhanced with Query Named Entity Recognition. This thesis proposes a practical three-phases QNER process and implements it on top of the leading open source search engine Elasticsearch. A stateless search application was built, benefiting from QNER process by using it for query scoping. The outcome of the project is a performant, scalable search architecture enhanced with query understanding capability.
This thesis focuses on the query understanding component of the search – the key to unlock next level of search relevance. Query understanding is an active area of research, giving rises to different techniques that aim at understanding the search intent behind the search query.
Among many tasks of query understanding, Query Named Entity Recognition (QNER) aims to decode user intent by identifying and classifying query segments of the search queries. The QNER process is the enablement behind many query transformations tasks such as query scoping, query relaxation, query expansion. In addition, it will simplify the rest of informational retrieval process and open the opportunities for advanced features such as search suggestion, personalization, and recommendation.
The objective of this thesis is to build search system for ecommerce enhanced with Query Named Entity Recognition. This thesis proposes a practical three-phases QNER process and implements it on top of the leading open source search engine Elasticsearch. A stateless search application was built, benefiting from QNER process by using it for query scoping. The outcome of the project is a performant, scalable search architecture enhanced with query understanding capability.