Design, implement, and evaluate a local-news chatbot for civic engagement

Thesis Abstract

Local news ecosystems increasingly rely on conversational interfaces to bridge gaps between residents and municipal information, yet access barriers and disengagement persist among underserved communities. This study addresses the problem of low participatory engagement in local governance despite abundant local-news content by designing, implementing, and evaluating a local-news chatbot intended to enhance civic participation. The aim is to develop a scalable chatbot that delivers timely local-news summaries, explains civic processes, and prompts user-appropriate actions (e.g., attending town-hall meetings, contacting representatives). Specific objectives include (1) designing a multilingual, accessible chatbot grounded in user-centered design and the normative theories of civic media and information diffusion; (2) implementing an end-to-end system integrating RSS/news APIs, a dialogue manager, and an sentiment-aware recommender to tailor content to user profiles; (3) evaluating usability, engagement, and informational outcomes among urban residents across five neighborhoods; and (4) assessing the impact on civic participation intentions and actual participation metrics over a 12-week trial. The study employs a mixed-methods research design anchored in design-research methodology and formative evaluation. The population comprises local news readers aged 18–65 within a mid-sized metropolitan area. A stratified sample of 360 residents is drawn across five neighborhoods, with 72 participants per neighborhood, selected to ensure diversity in language, age, and digital-literacy levels. Data collection instruments include (a) a validated usability questionnaire (System Usability Scale, SUS) and a bespoke engagement scale, (b) in-depth interviews with 40 participants to explore perceived barriers and enablers, (c) app interaction logs capturing metrics such as session length, frequency, feature usage, and click-through rates to civic-action prompts, and (d) pre- and post-trial surveys measuring civic participation intentions and actual actions (e.g., event attendance, contacting officials). Data analysis employs quantitative techniques—descriptive statistics, multivariate regression to identify predictors of engagement, ANOVA to compare across neighborhoods, and time-series analysis on participation metrics—and qualitative thematic analysis guided by Braun and Clarke to elucidate user experiences and contextual factors. The analytical framework situates the chatbot within the theory of Information Diffusion (Rogers) and the Civic Media framework, complemented by the Technology Acceptance Model (TAM) to interpret adoption dynamics. A conceptual model synthesizes user factors, chatbot features, and civic outcomes. Key expected findings include higher usability scores (SUS 75+), statistically significant increases in engagement indicators (session frequency, feature adoption) over the trial period, and positive shifts in civic-participation intentions and actions among users with higher digital literacy and prior civic engagement. It is anticipated that personalized content, proactive prompts, and multilingual support will correlate with greater engagement and more frequent exposure to local-government information, thereby increasing trust and willingness to participate in civic processes. The study also expects nuanced differences by neighborhood, elucidating how socio-demographic variables mediate chatbot effectiveness and suggesting design adaptations to mitigate disparities. The study contributes to knowledge by operationalizing a locally tailored, evidence-based chatbot design for civic engagement, providing a replicable blueprint for urban contexts and offering empirical evidence on how conversational interfaces influence local-news utilization and civic action. It advances theory by integrating Information Diffusion and Civic Media perspectives with user-acceptance constructs in a real-world civic-information system, and by demonstrating how language accessibility and personalization affect participation outcomes. Practical implications include guidelines for local news organizations and municipal agencies on deploying chatbots to improve accessibility to civic information, strategies for content curation and dialogue management that balance neutrality with action prompts, and governance considerations for data privacy and inclusivity. Based on findings, the study recommends iterative deployment with periodic user-testing, expansion to additional languages, integration with existing municipal platforms, and policy considerations to ensure equitable access to digital civic resources. The main conclusion posits that a well-designed local-news chatbot can modestly but measurably enhance civic engagement by lowering informational and logistical barriers, provided that content is accessible, local-contextualized, and aligned with residents’ civic needs; recommended actions include sustained maintenance, continuous user feedback loops, and partnerships with community organizations to broaden reach and impact.