Psych Literature Search

Problem

Literature reviews often stall at data cleaning: PubMed and Semantic Scholar return different schemas, while abstracts, DOIs, citation counts, variables, and measurement tools are scattered across sources. Extracting IVs, DVs, scale names, and scoring methods from fragmented records is hard to make auditable.

How the skill works

For a research topic, it runs a structured workflow:

• Parameter setup: topic, date range, and target paper count; optional WoS / Springer keys are not required for the main path.
• Parallel retrieval: queries PubMed via NCBI E-utilities and Semantic Scholar via REST API, both without API keys, and extracts title, abstract, doi, keywords, and citations. The scripts include rate limiting for keyless Semantic Scholar calls.
• Deduplication and ranking: merges records by DOI, then sorts by citation count and publication year.
• Structured output: produces a master table with linked title, keywords, original abstract, Chinese abstract, research variables, conceptual definitions, operational definitions, journal level, and citation count; an APA 7 list can be added when needed.

Boundaries

It is limited to metadata and abstracts, not full-text PDFs. Missing abstracts are marked [摘要不可获取]. If variables or definitions are not explicit, the skill infers them from the abstract and marks them (推断). Chinese topics are translated into English for retrieval, with the translated term noted; psychology-focused queries can add MeSH or fieldsOfStudy:Psychology filters.