An automated information extraction system from the knowledge graph based annual financial reports.

Area of Science:

• Computer Science
• Information Science
• Financial Informatics

Background:

• Annual financial reports contain crucial investment data but are unstructured and use varied terminology, hindering analysis.
• Manual data extraction and querying are complex and time-consuming for stakeholders.
• Existing digital systems struggle with the heterogeneity of financial report data.

Purpose of the Study:

• To develop a semantic web-based solution for automated information extraction from financial reports.
• To standardize financial terminologies and address semantic differences across institutions.
• To present extracted financial data in a queryable knowledge graph format.

Main Methods:

• Utilized an ontological approach for terminology standardization.
• Implemented semantic data extraction techniques to identify common semantics.
• Constructed a financial knowledge graph to represent and query the extracted information.

Main Results:

• The knowledge graph effectively makes financial information understandable and queryable.
• Demonstrated the utility of the financial knowledge graph in search engines, recommender systems, and Q-A systems.
• Achieved high precision and recall in competency question evaluations on bank datasets.

Conclusions:

• The proposed semantic web solution successfully addresses challenges in extracting and querying financial data.
• The financial knowledge graph enhances data accessibility and supports applications like financial storytelling.
• This approach offers a robust method for standardizing and utilizing information from diverse financial reports.