Outreach

At Mexlef 2022, scientists review advances in automatic language processing

Santa María Tonantzintla, Pue., October 7th, 2022. In order to learn about the most recent advances in research on the subject of automatic language processing, on October 6th and 7th, Mexican scientists met in Cholula in the Mexlef 2022 forum.

Mexlef 2022 was organized by the National Institute of Astrophysics, Optics and Electronics (INAOE), the Center for Scientific Research and Higher Education of Ensenada, Baja California (CICESE), the Engineering Institute of UNAM and the National Council for Science and Technology (Conacyt).

The analysis of information in digital media is a research topic of great interest and, despite current advances, there are still important challenges. At Mexlef 2022, the results of three tasks proposed by the Mexican scientific community to address some of these challenges and with special interest in the Spanish spoken in our country were presented.

In addition, the Mexlef 2022 program included keynote speeches and networking talks. The forum began yesterday with the conference "Mexican-style disinformation: political conditions for fertile ground for deception," with Dr. Martín Echeverría, from BUAP. Today, the work session began with the talk "Symanto, helping to create emotional connections with people", by Dr. Francisco Manuel Rangel Pardo, from Symanto Research.

Likewise, during these two days the results of the three tasks proposed by the scientific community have been presented: PAR-MEX (Paraphrase Identification in Mexican Spanish), Res-MEX (Recommendation System, Sentiment Analysis and Covid Semaphore Prediction for Mexican Tourist Texts) and DA-VINCIS (Detection of Aggressive and Violent Incidents from Social Media in Spanish).

It is worth mentioning that, as in previous years, the efforts to present evaluation tasks in Mexican Spanish are carried out in coordination with the IberLEF campaign. The three tasks this year were carried out during the period from March to July of this year under the framework of IberLEF.