arXiv subjects

Thiago H Silva

Publications and source records attributed to Thiago H Silva.

14 recordsLinked to original sources

arXiv2026

Persona Prompting in Multimodal Urban Perception: Descriptive Convergence and Interpretive Variation

This study examines how persona prompting shapes language generated by two multimodal large language models in urban perception, a setting for examining subjective interpretations of shared visual evidence. We organize outputs into three functional levels: descriptive grounding (captions), intermediate semantic layer (perception tags), and interpretive framing (justifications). Using approximately 60,000 persona-conditioned annotations per model from Qwen3-VL-8B and Gemma-4-E4B-it, we find that captions converge strongly across persona profiles and show only small attribute-associated differences. Justifications vary substantially more: economic status produces the largest difference in both models, with political orientation and personality also prominent. Paired image-level comparisons confirm larger justification than caption differences for these three attributes. For perception tags, personas sharing the same attribute level produce more similar tag sets than personas with different attribute levels, with the largest separation observed for economic status. Exploratory topic analysis further reveals persona-specific evaluative emphasis. Across models, profile-pair similarity patterns are strongly correlated for all three output types, although agreement is lowest for justifications. Overall, persona prompting affects interpretive framing more strongly than descriptive grounding.

cs.CL↗

arXiv2026

CityBehavEx: A Scalable and Empirically Validated LLM-Assisted Urban Simulation Platform

Recent LLM-based multi-agent urban simulators can generate semantically rich city routines, but they remain costly to scale and are often weakly validated against empirical mobility patterns. We present CityBehavEx, an interactive LLM-assisted urban simulation platform that scales to city-size populations, exposes agent behavior for inspection, supports empirical validation, and generates mobility patterns that better match real-world spatial, temporal, and semantic distributions. Instead of invoking large language models for every agent action, CityBehavEx combines established human mobility models with fine-tuned cross-encoders that estimate semantic alignment between agent profiles, schedules, and activity transitions. This design enables large-scale simulations, as demonstrated in a case study of 100,000 agents over 75 days in under one hour on a single consumer GPU. The platform allows users to define simulation regions, launch experiments, inspect trajectories and activity traces, debug unrealistic behaviors, and validate generated routines against real-world mobility, time-use, and semantic metrics.

cs.CL↗

arXiv2026

Stable Behavior, Limited Variation: Persona Validity in LLM Agents for Urban Sentiment Perception

Large Language Models (LLMs) are increasingly used as proxies for human perception in urban analysis, yet it remains unclear whether persona prompting produces meaningful and reproducible behavioral diversity. We investigate whether distinct personas influence urban sentiment judgments generated by multimodal LLMs. Using a factorial set of personas spanning gender, economic status, political orientation, and personality, we instantiate multiple agents per persona to evaluate urban scene images from the PerceptSent dataset and assess both within-persona consistency and cross-persona variation. Results show strong convergence among agents sharing a persona, indicating stable and reproducible behavior. However, cross-persona differentiation is limited: economic status and personality induce statistically detectable but practically modest variation, while gender shows no measurable effect and political orientation only negligible impact. Agents also exhibit an extremity bias, collapsing intermediate sentiment categories common in human annotations. As a result, performance remains strong on coarse-grained polarity tasks but degrades as sentiment resolution increases, suggesting that simple label-based persona prompting does not capture fine-grained perceptual judgments. To isolate the contribution of persona conditioning, we additionally evaluate the same model without personas. Surprisingly, the no-persona model sometimes matches or exceeds persona-conditioned agreement with human labels across all task variants, suggesting that simple label-based persona prompting may add limited annotation value in this setting.

cs.CL↗

arXiv2026

Grandes Modelos de Linguagem Multimodais (MLLMs): Da Teoria à Prática

Multimodal Large Language Models (MLLMs) combine the natural language understanding and generation capabilities of LLMs with perception skills in modalities such as image and audio, representing a key advancement in contemporary AI. This chapter presents the main fundamentals of MLLMs and emblematic models. Practical techniques for preprocessing, prompt engineering, and building multimodal pipelines with LangChain and LangGraph are also explored. For further practical study, supplementary material is publicly available online: https://github.com/neemiasbsilva/MLLMs-Teoria-e-Pratica. Finally, the chapter discusses the challenges and highlights promising trends.

cs.CL↗

arXiv2025

A Relational Model of Neighborhood Mobility: The Role of Amenities and Cultural Alignment

Why are some neighborhoods strongly connected while others remain isolated? Although standard explanations focus on demographics, economics, and geography, movement across the city may also depend on cultural styles and amenity mix. This study proposes a relational, cross-national model in which local culture and amenity mix alignment creates a "soft infrastructure" of urban mobility, i.e., symbolic cues and functional features that shape expectations about the character of places. Using ~650 million Google Places reviews to measure co-visitation between U.S. ZIP codes and ~30 million Canadian change-of-address to track residential mobility, results show that neighborhoods with similar cultural styles and amenities are significantly more connected. These effects persist even after controlling for race, income, education, politics, housing costs, and distance. Urban cohesion and segregation depend not only on who lives where or how far apart neighborhoods are, but on the shared cultural and material ecologies that structure movement across the city.

cs.SI↗

arXiv2025

CityHood: An Explainable Travel Recommender System for Cities and Neighborhoods

We present CityHood, an interactive and explainable recommendation system that suggests cities and neighborhoods based on users' areas of interest. The system models user interests leveraging large-scale Google Places reviews enriched with geographic, socio-demographic, political, and cultural indicators. It provides personalized recommendations at city (Core-Based Statistical Areas - CBSAs) and neighborhood (ZIP code) levels, supported by an explainable technique (LIME) and natural-language explanations. Users can explore recommendations based on their stated preferences and inspect the reasoning behind each suggestion through a visual interface. The demo illustrates how spatial similarity, cultural alignment, and interest understanding can be used to make travel recommendations transparent and engaging. This work bridges gaps in location-based recommendation by combining a kind of interest modeling, multi-scale analysis, and explainability in a user-facing system.

cs.IR↗

arXiv2024

Using Graph Neural Networks to Predict Local Culture

Urban research has long recognized that neighbourhoods are dynamic and relational. However, lack of data, methodologies, and computer processing power have hampered a formal quantitative examination of neighbourhood relational dynamics. To make progress on this issue, this study proposes a graph neural network (GNN) approach that permits combining and evaluating multiple sources of information about internal characteristics of neighbourhoods, their past characteristics, and flows of groups among them, potentially providing greater expressive power in predictive models. By exploring a public large-scale dataset from Yelp, we show the potential of our approach for considering structural connectedness in predicting neighbourhood attributes, specifically to predict local culture. Results are promising from a substantive and methodologically point of view. Substantively, we find that either local area information (e.g. area demographics) or group profiles (tastes of Yelp reviewers) give the best results in predicting local culture, and they are nearly equivalent in all studied cases. Methodologically, exploring group profiles could be a helpful alternative where finding local information for specific areas is challenging, since they can be extracted automatically from many forms of online data. Thus, our approach could empower researchers and policy-makers to use a range of data sources when other local area information is lacking.

cs.LG↗

arXiv2024

Bubble reachers and uncivil discourse in polarized online public sphere

Early optimism saw possibilities for social media to renew democratic discourse, marked by hopes for individuals from diverse backgrounds to find opportunities to learn from and interact with others different from themselves. This optimism quickly waned as social media seemed to breed ideological homophily marked by "filter bubble" or "echo chambers." A typical response to the sense of fragmentation has been to encourage exposure to more cross-partisan sources of information. But do outlets that reach across partisan lines in fact generate more civil discourse? And does the civility of discourse hosted by such outlets vary depending on the political context in which they operate? To answer these questions, we identified bubble reachers, users who distribute content that reaches other users with diverse political opinions in recent presidential elections in Brazil, where populism has deep roots in the political culture, and Canada, where the political culture is comparatively moderate. Given that background, this research studies unexplored properties of content shared by bubble reachers, specifically the quality of conversations and comments it generates. We examine how ideologically neutral bubble reachers differ from ideologically partisan accounts in the level of uncivil discourse they provoke, and explore how this varies in the context of the two countries considered. Our results suggest that while ideologically neutral bubble reachers support less uncivil discourse in Canada, the opposite relationship holds in Brazil. Even non-political content by ideologically neutral bubble reachers elicits a considerable amount of uncivil discourse in Brazil. This indicates that bubble reaching and incivility are moderated by the national political context. Our results complicate the simple hypothesis of a universal impact of neutral bubble reachers across contexts.

cs.SI↗

arXiv2022

Changing the Scene: applying four models of social evolution to the scenescape

This paper elaborates a multi-model approach to studying how local scenes change. We refer to this as the "4 D's" of scene change: development, differentiation, defense, and diffusion. Each posits somewhat distinct change processes, and has its own tradition of theory and empirical research, which we briefly review. After summarizing some major trends in scenes and amenities in the US context, for each change model, we present some initial findings, discussing data and methods throughout. Our overall goal is to point toward new research arcs on change models of scenes, and to give some clear examples and directions for how to think about and collect data to understand what makes some scenes change, others not, why, and in what directions.

cs.SI↗

arXiv2022

Reaching the bubble may not be enough: news media role in online political polarization

Politics in different countries show diverse degrees of polarization, which tends to be stronger on social media, given how easy it became to connect and engage with like-minded individuals on the web. A way of reducing polarization would be by distributing cross-partisan news among individuals with distinct political orientations, i.e., ``reaching the bubbles''. This study investigates whether this holds in the context of nationwide elections in Brazil and Canada. We collected politics-related tweets shared during the 2018 Brazilian presidential election and the 2019 Canadian federal election. Next, we proposed an updated centrality metric that enables identifying highly central bubble reachers, nodes that can distribute content among users with diverging political opinions - a fundamental metric for the proposed study. After that, we analyzed how users engage with news content shared by bubble reachers, its source, and its topics, considering its political orientation. Among other results, we found that, even though news media disseminate content that interests different sides of the political spectrum, users tend to engage considerably more with content that aligns with their political orientation, regardless of the topic.

cs.SI↗

arXiv2020

Automatic Extraction of Urban Outdoor Perception from Geolocated Free-Texts

The automatic extraction of urban perception shared by people on location-based social networks (LBSNs) is an important multidisciplinary research goal. One of the reasons is because it facilitates the understanding of the intrinsic characteristics of urban areas in a scalable way, helping to leverage new services. However, content shared on LBSNs is diverse, encompassing several topics, such as politics, sports, culture, religion, and urban perceptions, making the task of content extraction regarding a particular topic very challenging. Considering free-text messages shared on LBSNs, we propose an automatic and generic approach to extract people's perceptions. For that, our approach explores opinions that are spatial-temporal and semantically similar. We exemplify our approach in the context of urban outdoor areas in Chicago, New York City and London. Studying those areas, we found evidence that LBSN data brings valuable information about urban regions. To analyze and validate our outcomes, we conducted a temporal analysis to measure the results' robustness over time. We show that our approach can be helpful to better understand urban areas considering different perspectives. We also conducted a comparative analysis based on a public dataset, which contains volunteers' perceptions regarding urban areas expressed in a controlled experiment. We observe that both results yield a very similar level of agreement.

cs.SI↗

arXiv2020

Uncovering Spatiotemporal and Semantic Aspects of Tourists Mobility Using Social Sensing

Tourism favors more economic activities, employment, revenues and plays a significant role in development; thus, the improvement of this activity is a strategic task. In this work, we show how social sensing can be used to understand the key characteristics of the behavior of tourists and residents. We observe distinct behavioral patterns in those classes, considering the spatial and temporal dimensions, where cultural and regional aspects might play an important role. Besides, we investigate how tourists move and the factors that influence their movements in London, New York, Rio de Janeiro and Tokyo. In addition, we propose a new approach based on a topic model that enables the automatic identification of mobility pattern themes, ultimately leading to a better understanding of users' profiles. The applicability of our results is broad, helping to provide better applications and services in the tourism segment.

cs.SI↗

arXiv2017

Gender Matters! Analyzing Global Cultural Gender Preferences for Venues Using Social Sensing

Gender differences is a phenomenon around the world actively researched by social scientists. Traditionally, the data used to support such studies is manually obtained, often through surveys with volunteers. However, due to their inherent high costs because of manual steps, such traditional methods do not quickly scale to large-size studies. We here investigate a particular aspect of gender differences: preferences for venues. To that end we explore the use of check-in data collected from Foursquare to estimate cultural gender preferences for venues in the physical world. For that, we first demonstrate that by analyzing the check-in data in various regions of the world we can find significant differences in preferences for specific venues between gender groups. Some of these significant differences reflect well-known cultural patterns. Moreover, we also gathered evidence that our methodology offers useful information about gender preference for venues in a given region in the real world. This suggests that gender and venue preferences observed may not be independent. Our results suggests that our proposed methodology could be a promising tool to support studies on gender preferences for venues at different spatial granularities around the world, being faster and cheaper than traditional methods, besides quickly capturing changes in the real world.

cs.SI↗

arXiv2014

You are What you Eat (and Drink): Identifying Cultural Boundaries by Analyzing Food & Drink Habits in Foursquare

Food and drink are two of the most basic needs of human beings. However, as society evolved, food and drink became also a strong cultural aspect, being able to describe strong differences among people. Traditional methods used to analyze cross-cultural differences are mainly based on surveys and, for this reason, they are very difficult to represent a significant statistical sample at a global scale. In this paper, we propose a new methodology to identify cultural boundaries and similarities across populations at different scales based on the analysis of Foursquare check-ins. This approach might be useful not only for economic purposes, but also to support existing and novel marketing and social applications. Our methodology consists of the following steps. First, we map food and drink related check-ins extracted from Foursquare into users' cultural preferences. Second, we identify particular individual preferences, such as the taste for a certain type of food or drink, e.g., pizza or sake, as well as temporal habits, such as the time and day of the week when an individual goes to a restaurant or a bar. Third, we show how to analyze this information to assess the cultural distance between two countries, cities or even areas of a city. Fourth, we apply a simple clustering technique, using this cultural distance measure, to draw cultural boundaries across countries, cities and regions.

cs.SI↗