Periodic Reporting for period 3 - DYMO (Dynamic dialogue modelling)
Reporting period: 2022-09-01 to 2024-02-29
With the prevalence of information technology in our daily lives, our ability to interact with machines in increasingly simplified and more human-like ways has become paramount. Information is becoming ever more abundant but our access to it is limited not least by technological restraints. Spoken dialogue systems address this issue by providing an intelligent speech interface that facilitates swift, human-like acquisition of information.
The advantages of speech interfaces are already evident from the rise of personal assistants such as Siri, Google Assistant, Cortana or Amazon Alexa. In these systems, however, the user is limited to a simple query, and the systems attempt to provide an answer within one or two turns of dialogue. To date, significant parts of these systems are rule-based and do not readily scale to changes in the domain of operation. Furthermore, rule-based systems can be brittle when speech recognition errors occur.
The vision of this project is to develop novel dialogue models that provide natural human-computer interaction beyond simple information-seeking dialogues and that continuously evolve as they are being used by exploiting both dialogue and non-dialogue data. Building such robust and intelligent spoken dialogue systems poses serious challenges in artificial intelligence and machine learning. This project will develop novel methods based not only on natural language processing and machine learning, but also topological data analysis to achieve these challenging goals.
The advantages of speech interfaces are already evident from the rise of personal assistants such as Siri, Google Assistant, Cortana or Amazon Alexa. In these systems, however, the user is limited to a simple query, and the systems attempt to provide an answer within one or two turns of dialogue. To date, significant parts of these systems are rule-based and do not readily scale to changes in the domain of operation. Furthermore, rule-based systems can be brittle when speech recognition errors occur.
The vision of this project is to develop novel dialogue models that provide natural human-computer interaction beyond simple information-seeking dialogues and that continuously evolve as they are being used by exploiting both dialogue and non-dialogue data. Building such robust and intelligent spoken dialogue systems poses serious challenges in artificial intelligence and machine learning. This project will develop novel methods based not only on natural language processing and machine learning, but also topological data analysis to achieve these challenging goals.
This project has proposed a novel, flexible and data driven way to define user models. These are essential for training and evaluation but also for personalisation abilities of the systems that we plan to study.
We also worked on novel feedback mechanisms and how they can be incorporated in the system. These are particularly important in the context of goal-directed dialogues that we address in this project.
Through our collaborations, we achieved significant results in the area of understanding. More fundamentally, we approached the problem of utilising and understanding word representations from a topological data analysis perspective.
We also worked on novel feedback mechanisms and how they can be incorporated in the system. These are particularly important in the context of goal-directed dialogues that we address in this project.
Through our collaborations, we achieved significant results in the area of understanding. More fundamentally, we approached the problem of utilising and understanding word representations from a topological data analysis perspective.
All our research outputs resulted in improvements over the state of the art, and in particular:
- Data-driven user modelling
- Information gain for reward modelling
- Independence and uncertainty modelling in tracking
- Topological analysis of word representations
We plan to further develop the continual learning framework to mimic the lifespan of an evolving dialogue system across multiple domains. We aim to develop more sophisticated measures of feedback including emotional ones. We will make user models more human-like and investigate personalisation abilities of the system. We will invest significant efforts to further investigate the use of topological data analysis for dynamic ontology construction.
- Data-driven user modelling
- Information gain for reward modelling
- Independence and uncertainty modelling in tracking
- Topological analysis of word representations
We plan to further develop the continual learning framework to mimic the lifespan of an evolving dialogue system across multiple domains. We aim to develop more sophisticated measures of feedback including emotional ones. We will make user models more human-like and investigate personalisation abilities of the system. We will invest significant efforts to further investigate the use of topological data analysis for dynamic ontology construction.