Jump to Content
Felipe Goldstein

Felipe Goldstein

Research Areas

Authored Publications
results

Publications

Years

Research Areas

Teams

Sort By
  • Title
  • Title, descending
  • Year
  • Year, descending
    Recognizing Multimodal Entailment (tutorial at ACL 2021)
    Afsaneh Hajiamin Shirazi
    Arjun Gopalan
    Arsha Nagrani
    Blaž Bratanič
    Cesar Ilharco Magalhaes
    Christina Liu
    Felipe Ferreira (Goldstein)‎
    Gabriel Fedrigo Barcik
    Georg Fritz Osang
    Jared Frank
    Lucas Smaira
    Qin Cao
    Ricardo Abasolo Marino
    Roma Patel
    Thomas Leung
    Vaiva Imbrasaite
    (2021) (to appear)
    Preview abstract How information is created, shared and consumed has changed rapidly in recent decades, in part thanks to new social platforms and technologies on the web. With ever-larger amounts of unstructured and limited labels, organizing and reconciling information from different sources and modalities is a central challenge in machine learning. This cutting-edge tutorial aims to introduce the multimodal entailment task, which can be useful for detecting semantic alignments when a single modality alone does not suffice for a whole content understanding. Starting with a brief overview of natural language processing, computer vision, structured data and neural graph learning, we lay the foundations for the multimodal sections to follow. We then discuss recent multimodal learning literature covering visual, audio and language streams, and explore case studies focusing on tasks which require fine-grained understanding of visual and linguistic semantics question answering, veracity and hatred classification. Finally, we introduce a new dataset for recognizing multimodal entailment, exploring it in a hands-on collaborative section. Overall, this tutorial gives an overview of multimodal learning, introduces a multimodal entailment dataset, and encourages future research in the topic. View details
    Extracting Unambiguous Keywords from Microposts Using Web and Query Logs Data
    Davi Reis
    Felipe Goldstein
    Frederico Quintao
    Making sense of Microposts (at WWW 2012)
    Preview abstract In the recent years, a new form of content type has become ubiquitous in the web. These are small and noisy text snippets, created by users of social networks such as Twitter and Facebook. The full interpretation of those microposts by machines impose tremendous challenges, since they strongly rely on context. In this paper we propose a task which is much simpler than full interpretation of microposts: we aim to build classification systems to detect keywords that unambiguously refer to a single dominant concept, even when taken out of context. For example, in the context of this task, apple would be classified as ambiguous whereas microsoft would not. The contribution of this work is twofold. First, we formalize this novel classification task that can be directly applied for extracting information from microposts. Second, we show how high precision classifiers for this problem can be built out of Web data and search engine logs, combining traditional information retrieval metrics, such as inverted document frequency, and new ones derived from search query logs. Finally, we have proposed and evaluated relevant applications for these classifiers, which were able to meet precision ≥ 72% and recall ≥ 56% on unambiguous keyword extraction from microposts. We also compare those results with closely related systems, none of which could outperform those numbers. View details
    Search on Google Scholar