"Dense Template Retrieval for Customer Support", Tiago Mesquita (Zendesk)

[Diogo Pernes‏]

Dear all,

We are excited to announce that after almost three years of remote events, the Priberam Machine Learning Lunch Seminars are back in an in-person format!

To kick off our series, we're hosting a presentation by Tiago Mesquita, a Machine Learning Scientist at Zendesk. Tiago will be discussing his work on a dense retrieval framework for the customer support scenario.

The event will take place on Tuesday next week, March 7th, at 1 PM (WET) in Instituto Superior Técnico (room PA2). As in past in-person editions, we'll provide lunch bags, so please come hungry!

To learn more about the event and to register (mandatory if you'd like to attend), please follow the link below:
https://www.eventbrite.pt/e/dense-template-retrieval-for-customer-support-tickets-564740031637

We hope to see you all there, and we look forward to reconnecting in person.

Kind regards,
Diogo Pernes

Priberam Labs
http://labs.priberam.com/

Priberam is hiring!

If you are interested in working with us please consult the available positions at priberam.com/careers.

PRIBERAM SEMINARS

__________________________________________________

Priberam Machine Learning Lunch Seminar
Speaker: Tiago Mesquita (Zendesk)
Venue: Instituto Superior Técnico (room PA2)

Date: Tuesday, March 7, 2023
Time: 1 PM 

Title:

Dense Template Retrieval for Customer Support

Abstract:

Templated answers are used extensively in customer support scenarios, providing an efficient way to cover a plethora of topics, with an easily maintainable collection of templates. However, the number of templates is often too high for an agent to manually search. Automatically suggesting the correct template for a given question can thus improve the service efficiency, reducing costs and leading to a better customer satisfaction. In this work, we propose a dense retrieval framework for the customer support scenario, adapting a standard in-batch negatives technique to support unpaired sampling of queries and templates. We also propose a novel loss that extends the typical query-centric similarity, exploiting other similarity relations in the training data. Experiments show that our approach achieves considerable improvements, in terms of performance and training speed, over more standard dense retrieval methods. This includes methods such as DPR, and also ablated versions of the proposed approach.

Short Bio:

Tiago Mesquita is a Machine Learning Scientist at Zendesk. He holds a master's degree (MSc) in Computer Science and Engineering obtained from Instituto Superior Tecnico in 2021. His current line of research is centered around machine learning methods applied to customer support. Tiago´s other research interests include Natural Language Processing and Computer Vision.

More info: http://labs.priberam.com/Academia-Partnerships/Seminars.aspx

Eventbrite:

https://www.eventbrite.pt/e/dense-template-retrieval-for-customer-support-tickets-564740031637

[Diogo Pernes‏]

Dear all,

The first Priberam ML Lunch Seminar of the year will take place tomorrow, March 7th, at 1 PM (WET) in Instituto Superior Técnico (room PA2). As in past in-person editions, we'll provide lunch bags, so please come hungry!

To kick off our series, we're hosting a presentation by Tiago Mesquita, a Machine Learning Scientist at Zendesk. Tiago will be discussing his work on a dense retrieval framework for the customer support scenario.

Senior Research Scientist at Priberam

Invited Assistant Professor at the University of Porto