Experimenting with UD Adaptation of an Unsupervised Rule-based Approach for Sentiment Analysis of Mexican Tourist Texts

This paper summarizes the results of experimenting with Universal Dependencies (UD) adaptation of an Unsupervised, Compositional and Recursive (UCR) rule-based approach for Sentiment Analysis (SA) submitted to the Shared Task at Rest-Mex 2023 (Team Olga/LyS-SALSA) (within the IberLEF 2023 conference). By using basic syntactic rules such as rules of modification and negation applied on words from sentiment dictionaries, our approach exploits some advantages of an unsupervised method for SA: (1) interpretability and explainability of SA, (2) robustness across datasets, languages and domains and (3) usability by non-experts in NLP. We compare our approach with other unsupervised approaches of SA that in contrast to our UCR rule-based approach use simple heuristic rules to deal with negation and modification. Our results show a considerable improvement over these approaches. We discuss future improvements of our results by using modality features as another shifting rule of polarity and word disambiguation techniques to identify the right sentiment words.

Geolocation differences of language use in urban areas

The explosion in the availability of natural language data in the era of social media has given rise to a host of applications such as sentiment analysis and opinion mining. Simultaneously, the growing availability of precise geolocation information is enabling visualization of global phenomena such as environmental changes and disease propagation. Opportunities for tracking spatial variations in language use, however, have largely been overlooked, especially on small spatial scales. Here we explore the use of Twitter data with precise geolocation information to resolve spatial variations in language use on an urban scale down to single city blocks. We identify several categories of language tokens likely to show distinctive patterns of use and develop quantitative methods to visualize the spatial distributions associated with these patterns. Our analysis concentrates on comparison of contrasting pairs of Tweet distributions from the same category, each defined by a set of tokens. Our work shows that analysis of small-scale variations can provide unique information on correlations between language use and social context which are highly valuable to a wide range of fields from linguistic science and commercial advertising to social services.