Five focused prototypes explore journal fit, semantic discovery, author networks, topic movement and journal-level publishing patterns. Each one turns the same bounded content foundation into a different editorial or portfolio question.
Open a live experience, then follow its documentation links to see the request path, data boundaries and offline build.
Paste a title and abstract to compare journal similarity, inspect related-article evidence and place recommendations on a two-dimensional portfolio map.
Run vector retrieval across the portfolio, filter by journal and month, compare reranking choices, inspect scores and synthesize a grounded answer from the returned evidence.
Explore 32,981 papers and 164,838 disambiguated author identities through four question-shaped views, with the graph analysis completed before deployment.
Browse 500 materialised topic summaries, inspect longitudinal series and article evidence, and compare two to four topics across the portfolio.
Explore journal publishing breadth, article volume and topic drift, then compare other journals showing similar topic momentum using precomputed portfolio evidence.