In-Context Learning User Simulators for Task-Oriented Dialogs
@ LLM4AI’23: Workshop on Foundations and Applications in Large-scale
AI Models -Pre-training, Fine-tuning, and Prompt-based Learning co-located at KDD2023 (07/08/2023)
We present a novel application of large language models in user simulation for task-oriented dialog systems,
specifically focusing on an in-context learning approach.
By harnessing the power of these models, the proposed approach generates
diverse utterances based on user goals and limited dialog examples.
Unlike traditional simulators, this method eliminates the need for
labor-intensive rule definition or extensive annotated data,
making it more efficient and accessible.
[Slides]
BETOLD: A Task-Oriented Dialog Dataset for Breakdown Detection @ When creative AI meets conversational AI (co-located at COLING 2022) (12/11/2022)
Task-Oriented Dialog (TOD) systems often suffer from dialog breakdowns -
situations in which users cannot or do not want to proceed with the conversation.
Ideally TOD systems should be able to detect dialog breakdowns to prevent users
from quitting a conversation and to encourage them to interact with the system
again. In this paper, we present BETOLD, a privacy-preserving dataset for
breakdown detection. The dataset consists of user and system turns represented
by intents and entity annotations, derived from NLU and NLG dialog manager
components. We also propose an attention-based model that detects potential
breakdowns using these annotations, instead of the utterances’ text. This
approach achieves a comparable performance to the corresponding utterance-only
model, while ensuring data privacy.
[Slides]
Cross-lingual Contextualized Topic Models with Zero-shot Learning @ CLIC-it (01/07/2022)
Many data sets (e.g., reviews, forums, news, etc.) exist parallelly in multiple languages.
They all cover the same content, but the linguistic differences make it
impossible to use traditional, bag-of-word-based topic models.
Models have to be either single-language or suffer from a huge, but
extremely sparse vocabulary. Both issues can be addressed by transfer learning.
We introduce a zero-shot cross-lingual topic model.
Our model learns topics on one language (here, English),
and predicts them for unseen documents in different languages
(here, Italian, French, German, and Portuguese).
[Slides]
OCTIS 2.0: Optimizing and Comparing Topic Models in Italian Is Even Simpler @ CLIC-it (30/06/2022)
OCTIS is an open-source frame-work for training, evaluating and compar-
ing Topic Models. This tool uses single-objective Bayesian Optimization (BO) to
optimize the hyper-parameters of the models and thus guarantee a fairer
comparison. Yet, a single-objective approach disregards that a user may want
to simultaneously optimize multiple objectives. We therefore propose OCTIS 2.0:
the extension of OCTIS that addresses the problemof estimating the optimal hyper-parameter
configurations for a topic model using multi-objective BO. Moreover, we also
release and integrate two pre-processed Italian datasets, which can be easily used as
benchmarks for the Italian language.
[Slides]
Beyond the Bag Of Words: Text Analysis with Contextualized Topic Models @ NLP+CSS 201 Tutorials (22/11/2021)
Most topic models still use Bag-Of-Words (BoW) document representations as input.
These representations, though, disregard the syntactic and semantic relationships
among the words in a document, the two main linguistic avenues to coherent text.
Recently, pre-trained contextualized embeddings have enabled exciting new results
in several NLP tasks, mapping a sentence to a vector representation.
Contextualized Topic Models (CTM) combine contextualized embeddings with
neural topic models to increase the quality of the topics. Moreover, using
multilingual embeddings allows the model to learn topics in one language and
predict them for documents in unseen languages, thus addressing a task of
zero-shot cross-lingual topic modeling.
[Slides]
[Code]
[Video]
Beyond the Bag Of Words: Text Analysis with Contextualized Topic Models @ Universität Bern (12/11/2021)
Yet another tutorial on Topic Models, how to use them and evaluate them, with a focus on neural topic models.
[Slides]
Modeling Knowledge Incorporation into Topic Models and their Evaluation
@ EURECOM (17/06/2021)
Topic models are statistical methods that aim at extracting the themes, or "topics", from large collections of documents.
We may have some knowledge, associated with the documents (e.g. document labels, pre-trained representations) that can be exploited to improve the
quality of the resulting topics. In this talk, I will review different methods to incorporate knowledge into topic models. Moreover,
due to their stochastic and unsupervised nature, topic models are difficult to evaluate. Therefore, I will discuss the issues of their
evaluation and show how to guarantee a fairer comparison between the models. [Slides]
Natural Language Processing
and Topic Modeling Review @ AllianceBernstein (30/10/2020)
A review of the tremendous progress in the field of Natural Language Processing, including Language Models and Topic Models.
I also present Contextualized Topic Models, that get the best of both worlds. [Slides]