Recover 2018: 1st International Workshop on REbooting the COnVErsational Recommender https://recover18.wordpress.com/ Vancouver, Canada, October 5-8, 2018 |
Conference website | https://recover18.wordpress.com/ |
Submission link | https://easychair.org/conferences/?conf=recover2018 |
Abstract registration deadline | July 16, 2018 |
Submission deadline | July 16, 2018 |
===================================================================================================================
1st International Workshop on REbooting the COnVErsational Recommender Systems (Recover 2018):https://recover18.wordpress.com/
To be held in Vancouver (Canada). Co-located with ACM Conference on Recommender Systems 2018:https://recsys.acm.org/recsys18/
===================================================================================================================
Consider a recommender which engages with its user to help her to articulate her short- or longer-term preferences. Consider too a recommender which invites the user to express an opinion about tentative recommendations in order to guide the recommender in making further recommendations. In both cases, there is a cycle of interactions between the user and the recommender. We will refer to these as conversational recommenders, and they are the topic of this workshop.
Note that the phrase 'conversational' as defined here neither implies, nor excludes, recommenders that conduct dialogs in natural language. A conversational recommender might converse in natural language, but it may allow more constrained modes of user interaction too.
Research into conversational recommenders was a prominent strand in the late 1990s and early 2000s. Papers on preference elicitation through question-asking and on recommendation critiquing ("like this but cheaper") were common.
While research and development into conversational recommenders has never gone away, it has certainly been less prominent for a while.But this seems to be changing. There seems to be renewed interest in conversational recommenders.
It is this renewed interest that furnishes the rationale for holding this workshop at this time.
-------------------
Topics of Interest
-------------------
We will invite papers that pertain to the workshop theme including but not limited to:
design of conversational agents
critiquing in conversational recommenders
question-asking in conversational recommenders
explanations in conversational recommenders
active learning in conversational recommenders
user modelling for conversational recommenders
natural language interaction with recommenders
natural language processing for conversational recommenders
speech interfaces for conversational recommenders
dialogue management for conversational recommenders
UX design for conversational recommenders
conversation analysis for conversational recommenders
chatbots for conversational recommenders
conversational group recommenders
interaction methods in conversational recommenders
-------------------
Submissions
-------------------
Authors may submit fully-developed ideas and approaches as long papers (8 pages) and preliminary work as short papers (4 pages). For full details on the submission format and procedure, see below. Papers will be selected based on originality, quality, and ability to promote discussion. Accepted papers will be included in the workshop proceedings and published by CEUR. Extended versions of selected workshop papers may be included in a special journal issue (TBD). At least one author of each accepted paper must attend the workshop.
All submissions should be in English and should not have been published or submitted for publication elsewhere. Papers should be formatted in the ACM Proceedings Style (http://www.acm.org/sigs/publications/proceedings-templates) and submitted via EasyChair (https://easychair.org/conferences/?conf=recover2018).
Submissions will be published in the workshop proceedings (CEUR-WS).
-------------------
Shared Task
-------------------
In this edition of the workshop, we propose a shared task on Conversational Recommender Systems. In particular, relying on the dataset proposed in [1], we encourage participants to develop systems able to both recommend items and converse with the user.
The proposed dataset contains automatically created dialogues from two well-known datasets: MovieLens 1M and MovieTweetings. The generated dataset is in JSON format and is split in validation, training and testing. This dataset allows computing the performance of a conversational recommender in terms of precision and dialogue accuracy measured through the BLEU measure.
The dataset is available here:https://github.com/swapUniba/ConvRecSysDataset.
A description of a system that uses these datasets is reported in [2].
[1] Alessandro Suglia, Claudio Greco, Pierpaolo Basile, Giovanni Semeraro and Annalina Caputo. An automatic procedure for generating datasets for conversational recommender systems. In Proceedings of Dynamic Search for Complex Tasks-8th International Conference of the CLEF Association, CLEF.
[2] Claudio Greco, Alessandro Suglia, Pierpaolo Basile and Giovanni Semeraro. Converse-Et-Impera: Exploiting Deep Learning and Hierarchical Reinforcement Learning for Conversational Recommender Systems. In Proceedings of 16th International Conference of the Italian Association for Artificial Intelligence, 372-386, Springer 2017
------------------------
Important dates
------------------------
- Paper submission deadline: July 16th, 2018
- Author notification: July 31th, 2018
- Camera-ready version deadline: August 4th, 2018
- Workshop (at RecSys 2018): October 2-7, 2018