ProZe: Explainable and Prompt-Guided Zero-Shot Text Classification

July 2022, IEEE Internet Computing, Special issue on knowledge-infused learning for computational social systems.

To improve the state of the art on zero-shot text classification, we combine the explanatory power of a common-sense knowledge graph with the world knowledge contained in pretrained-language models by conditioning them on domain-specific prompts.

[ paper ]

Detecting COVID-19-related conspiracy theories in tweets

December 2021, 12th MediaEval Benchmarking Initiative for Multimedia Evaluation Workshop (MediaEval'2021), Online.

To tackle the problem of detecting COVID-19-related conspiracy theories in tweets, we used different approaches such as a combination of TFIDF and machine learning, transformer-based neural networks or Natural Language Inference.

[ code ]

Explainable zero-shot topic extraction using a common-sense knowledge graph

September 2021, 3rd Conference on Language, Data and Knowledge (LDK'2021), Zaragoza, Spain.

Wanna build a text classifier without any training data that can also explain its predictions? ZeSTE may be what you're looking for!

[ paper | code ]

And cut! Exploring textual representations for media content segmentation and alignment

June 2021, 2nd International Workshop on Data-driven Personalisation of Television (DataTV @ IMX'2021), Online.

In this work, we present an approach to content segmentation that leverages topical coherence, language modeling and word embeddings to detect change of topics.

[ paper ]

Named Entity Recognition as Graph Classification

June 2021, 18th Extended Semantic Web Conference (ESWC'2021 - Poster Track), Online.

Injecting real-world information (typically contained in Knowledge Graphs) and hand-crafted features into a pipeline for training end-to-end Natural Language Processing models is an open challenge. In this paper, we propose to approach the task of Named Entity Recognition, which is traditionally viewed as a Sequence Tagging problem, as a Graph Classification problem.

[ paper | code ]

Discovering interpretable topics by leveraging common sense knowledge

September 2021, 3rd Conference on Language, Data and Knowledge (LDK'2021)

How to make the results of topic modeling algorithms more understandable to humans? Try to add some common sense into the process :)

[ paper | code ]

Apples to Apples: A Systematic Evaluation of Topic Models

September 2021, 13th Conference on Recent Advances in NLP (RANLP'2021), Online

Topic Modeling Evaluation is an open problem in the Topic Modeling community. While the reliance on automatic evaluation remains more or less necessary to quickly assess the performance of a given topic model algorithm, there is no study that attempts to evaluate several algorithms in the literature given the same preprocessing, datasets, and metrics. That's what we did!

[ paper ]

ToModAPI: A Topic Modeling API to Train, Use and Compare Topic Models

November 2020, 2nd Workshop for NLP Open Source Software (NLP-OSS @ EMNLP'2020)}, Online.

This API is built to dynamically perform training, inference, and evaluation for different topic modeling techniques. The API grant common interfaces and command for accessing the different models, make easier to compare them.

[ paper | code ]

Stories of Love and Violence: Zero-shot interesting events classification for unsupervised TV series summarization

January 2023, Multimedia Systems Journal (2023).

Exploring Multimodality, Perplexity and Explainability forMemorability Prediction

December 2021, 12th MediaEval Benchmarking Initiative for Multimedia Evaluation Workshop (MediaEval'2021), Online

Character-based TV Series Summaries using keyword classification

December 2021, International Workshop on Video Retrieval Evaluation (TRECVID'2021), Online

[ code ]

Predicting Media Memorability with Audio, Video, and Text representation

December 2020, 11th MediaEval Benchmarking Initiative for Multimedia Evaluation Workshop (MediaEval'2020), Online

[ paper | code ]

Using Fan-Made Content, Subtitles and Face Recognition for Character-Centric Video Summarization

November 2020, the International Workshop on Video Retrieval Evaluation (TRECVID'2020), Online

[ paper | code ]

Combining Textual and Visual Modeling for Predicting Media Memorability

October 2019, 10th MediaEval Benchmarking Initiative for Multimedia Evaluation Workshop (MediaEval'2019), Sophia Antipolis, France.

[ paper | code ]

Modeling And Using The H2020 MeMAD Knowledge Graph (talk)

June 2019, The EBU Metadata Developer Network Workshop (EBU-MDN), Geneva, Switzerland

In the context of the European research project MeMAD (Methods for Managing Audiovisual Data), we face the challenge of modeling semantically audiovisual legacy metadata and results of automatic analysis from multiple partners and in an interoperable manner. In this talk, we will present an implementation of the EBU-CCDM/EBU Core data model for representing production and broadcasting information of TV and Radio programs provided by two industrial partners covering several channels. The resulting MeMAD knowledge graph provides metadata for more than 60K hours of audiovisual content, spanning multiple channels, audiovisual genres, themes and languages.

Improving Media Content Recommendation with Automatic Annotations

September 2021, the 3rd Edition of Knowledge-aware and Conversational Recommender Systems (KaRS @ RecSys'2021), Amsterdam, Netherlands

In this work, we study the potential of using off-the-shelf automatic annotation tools from the Information Extraction literature to improve recommendation performance without any extra cost of training, data collection or annotation.

[ paper | code ]