Food and Flavor Representation Learning

Companion Materials: Food and Flavor Representation Learning Food and Flavor Representation Learning: A Systematic Mapping Study and Research Agenda Purpose This package releases the operational artifacts behind the mapping study so that the screening, coding, and figure-generation decisions are auditable and the corpus can be reused as a starting point for subsequent reviews. Contents supplementary/├── README.md This file├── LICENSE CC BY 4.0 license text├── coding-guide.md Controlled vocabularies for scope,│ primary_topic, modalities, method_family├── corpus/│ ├── corpus.json All 323 screened entries with full coding│ ├── corpus.csv Same data in flat tabular form│ ├── quality-rubric.csv Six-dimension quality rubric for the│ │ 115 core-scope entries│ └── core.bib BibTeX entries for the 110 synthesis-set│ studies├── queries/│ ├── scopus.md Full unabbreviated query, syntax, date,│ │ retrieved-count for each database│ ├── sciencedirect-springer.md│ ├── ieee-xplore.md│ ├── acm-dl.md│ ├── arxiv.md│ └── pubmed.md└── figures/ ├── generate_figures.py Script that reproduces every figure in │ the manuscript from corpus.json └── data/ Derived count tables consumed by the figure script How to reproduce the figures cd supplementary/figurespython3 -m pip install matplotlib seabornpython3 generate_figures.py Outputs are written to `figures/output/` and match the PDFs included with the manuscript. How to re-execute the searches Each file under `queries/` records the exact Boolean expression, the date the query was executed, and the raw retrieved-count. Searches were run between January and February 2026; small drift in totals is expected for later re-executions because each database is continuously updated. How to interpret the corpus scope partitions the screened pool into core (representation contribution; synthesis set), adjacent (uses a representation as a component of a downstream application), `tangential` (cited only for historical, biological, or infrastructural context), and out_of_scope (excluded at full-text review) primary_topic and secondary_topics use the controlled vocabulary in coding-guide.md method_family and modalities are non-exclusive multi-valued fields synthesis_set is a boolean flag marking the 110 entries that contribute to the manuscript synthesis. The 115 scope == "core" entries are a superset: the 5 with synthesis_set == False were retained in the JSON for context but were not promoted into the manuscript taxonomy after final scope refinement. quality-rubric.csv covers all 115 core entries Changes v1.0 (2026-05): initial release accompanying manuscript submission.