Research and publications
Questions, methods, and programs
BLOCKMINE is built to support peer-reviewed research. This index lists the platform's research questions, its evaluation plan, and the research programs it hosts.
Research questions
- Can multi-source evidence retrieval identify unmet needs more precisely than single-source keyword search?
- Can transparent, evidence-conditioned ranking improve the usefulness of open problem discovery compared with generic LLM brainstorming?
- How does combining scientific publication trends, clinical trials, and population burden change healthcare gap identification?
- Can calibrated combinations of exploitation evidence, severity, and product applicability improve defensive vulnerability triage?
- Can publicly supportable dependency and disruption signals reveal supply chain fragility without falsely asserting private supplier relationships?
- Does a shared evidence graph produce useful relationships across opportunities and threats that would be missed by independent module search?
- How do freshness, missing data, source independence, and language or geographic coverage affect discovery quality?
Evaluation plan (summary)
- Versioned benchmark sets per module with negatives, borderline cases, stale and contradictory sources; temporal and entity-respecting splits.
- Two independent domain reviewers on an evaluation sample; agreement reported (Cohen's κ, Krippendorff's α, ICC) only when sufficient data exist.
- Per-module metrics: precision@k, nDCG, citation faithfulness, CVE join precision, version-match accuracy, graph edge precision, concentration calculation error.
- Baselines: single-source keyword retrieval; lexical + metadata; embeddings without graph; generic LLM without retrieval.
- Bootstrap intervals, sensitivity to score weights, coverage-bias audits by geography, language, and sector.
Research programs
Research program · AGGIES Lab
Blockchain Mining Research
Mining public blockchains for evidence. We crawl publicly accessible data across different public chains at scale and read it for signals of opportunity and threat, reporting only what the records support.
- 01 Public-Chain Data Collection
- 02 Adoption and Activity Signals
- 03 Provenance and Real-World Assets
- 04 Exploits and Contract Weaknesses
- 05 Concentration and Systemic Dependency
- 06 Ethics, Legality, and Reproducibility
Publications and working papers
- Journal articlePublished
Blockchain Applications for Cyber Liability Insurance
Kumar, S., Kocian, L., and Loo, L. (2024)
- Working paperIn progress
Enhancing Blockchain Enabled Resilience in the U.S. Beef Supply Chain
Veetil, V., Kaniyamattam, K., and Kumar, S. (2025)
Only completed or active work by the team is listed. Items marked In progress are not yet peer reviewed and may change in title, authorship, and venue.