Real-Time Multi-Stream Data Summarization for Enhanced Crisis Response and Situational Awareness

[202103002]

Title

CrisisFACTS: Real-Time Multi-Stream Data Summarization for Enhanced Crisis Response and Situational Awareness

Team Name

RescueRetrievers

Email

202103015@daiict.ac.in

Team Member 1 Name

Asish Joel Batha

Team Member 1 Id

202103015

Team Member 2 Name

Aadesh Minz

Team Member 2 Id

202103002

Team Member 3 Name

Raj Kariya

Team Member 3 Id

202103048

Team Member 4 Name

NIL

Team Member 4 Id

NIL

Category

Evaluation Track Problem

Problem Statement

The project aims to develop a system which enchances situational awareness during crisis by summarizing crucial real-time information from various data streams such as Twitter, Reddit, and Web News. The system will generate concise daily summaries of the most relevant facts, enabling local crisis response teams to access timely and relevant information, thereby improving disaster response coordination and decision-making.

Evaluation Strategy

The project uses the Rouge Score and BERT-Score to evaluate the system's performance in generating summaries.

Dataset

crisisFACTS - https://crisisfacts.github.io/

Resources

CrisisFACTS overview paper - https://crisisfacts.github.io/assets/pdf/crisisfacts2022.iscram2023.pdf

Lin, C.-Y. (2004). ROUGE: A Package for Automatic Evaluation of Summaries. Proceedings of the Workshop on Text Summarization Branches Out.

Zhang, T., Kishore, V., Wu, F., et al. (2019). BERTScore: Evaluating Text Generation with BERT. arXiv preprint: arXiv:1904.09675.

[parth126]

• Need more clarity on how the "real time" aspect will be simulated
• How will it be evaluated?