DialogueKitRec: A Library for Conversational Recommender Systems
Abstract
Conversational Recommender Systems (CRS) are made to have a dialogue with the user. The task is to make personalized recommendation, and the items a CRS recommends can be a reservation to a flight journey, or an item in an e-commerce store like a specific mobile phone. It may also just be what movie to watch this weekend.
This thesis aims to create a library named DialogueKitRec to simplify the creation of new Conversational Recommender Systems. This will allow any domain to use this template in creating the CRS. The domains may include e-commerce, music or movies.
We build DialogueKitRec on top of the already existing library DialogueKit. DialogueKit is a library for Conversational Information Access (CIA), and this system is made specifically for Natural Language Processing (NLP) and allowing dialogue with a CRS.
Finally, we show how DialogueKitRec can be used by refactoring MovieBot, this shows that it will be possible to create other CRS with DialogueKitRec as a base. Conversational Recommender Systems (CRS) are made to have a dialogue with the user. The task is to make personalized recommendation, and the items a CRS recommends can be a reservation to a flight journey, or an item in an e-commerce store like a specific mobile phone. It may also just be what movie to watch this weekend.
This thesis aims to create a library named DialogueKitRec to simplify the creation of new Conversational Recommender Systems. This will allow any domain to use this template in creating the CRS. The domains may include e-commerce, music or movies.
We build DialogueKitRec on top of the already existing library DialogueKit. DialogueKit is a library for Conversational Information Access (CIA), and this system is made specifically for Natural Language Processing (NLP) and allowing dialogue with a CRS.
Finally, we show how DialogueKitRec can be used by refactoring MovieBot, this shows that it will be possible to create other CRS with DialogueKitRec as a base.