Community Needs and Assets: A Computational Analysis of Community Conversations
This work addresses a gap in the literature for non-profits and government agencies by providing a tool to automate the analysis of community strengths and issues from social media, though it is incremental as it applies existing NLP methods to a new domain-specific dataset.
The paper tackles the challenge of manually analyzing social media conversations for community needs assessments by introducing a computational task to identify and categorize community needs and assets from conversational data, achieving an F1 score of 94% with their model compared to baselines of 49% and 61%.
A community needs assessment is a tool used by non-profits and government agencies to quantify the strengths and issues of a community, allowing them to allocate their resources better. Such approaches are transitioning towards leveraging social media conversations to analyze the needs of communities and the assets already present within them. However, manual analysis of exponentially increasing social media conversations is challenging. There is a gap in the present literature in computationally analyzing how community members discuss the strengths and needs of the community. To address this gap, we introduce the task of identifying, extracting, and categorizing community needs and assets from conversational data using sophisticated natural language processing methods. To facilitate this task, we introduce the first dataset about community needs and assets consisting of 3,511 conversations from Reddit, annotated using crowdsourced workers. Using this dataset, we evaluate an utterance-level classification model compared to sentiment classification and a popular large language model (in a zero-shot setting), where we find that our model outperforms both baselines at an F1 score of 94% compared to 49% and 61% respectively. Furthermore, we observe through our study that conversations about needs have negative sentiments and emotions, while conversations about assets focus on location and entities. The dataset is available at https://github.com/towhidabsar/CommunityNeeds.