MEMORA: Enterprise Knowledge Reasoning and Organizational Memory Platform
Organizations generate useful knowledge everyday in the form of project reports, technical documents,standard operating procedures, meeting records, emails,internal repositories, etc. The problem is not always a lack of information, it is often the work to find the right information and verify where it came from. This paper presents MEMORA, an Enterprise Knowledge Reasoning and Organizational Memory Platform, which aims at facilitating the retrieval and inspection of organizational knowledge. We now have a prototype with a Python Flask backend, semantic retrieval using the FAISS vector-search library, Cross-Encoder reranking, an Ask MEMORA query workflow, and presentation of evidence sources. We also built a visualization prototype, a knowledge graph interface implemented in React Flow. The paper describes the retrieval pipeline we implemented and proposes a controlled evaluation comparing FAISS retrieval with a two-stage FAISS and Cross-Encoder approach. It also discusses source traceability as a requirement distinct from retrieval relevance. However, as there are no verified query annotations and experiment logs available in the project records, this paper does not report numerical performance improvements. Persistent graph storage, graph-based retrieval, expert discovery, decision memory and knowledge-risk analysis are future extensions (rather than finished features).
Authors
- Vijay Kedar
- Chaitanya Jadhav
- Riddhita Tingare
- Purvaja Pawar
- Ashlesha Shinde
Institutions
- Dr. D. Y. Patil Medical College, Hospital and Research Centre (IN)
- DY Patil Deemed to be University (IN)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-18
- DOI
- https://doi.org/10.5281/zenodo.24342392
- Primary Topic
- Information Retrieval and Search Behavior
- Type
- preprint