CoolFace
Apppublic

ladidadigaard/sociology-search

sourceHugging Faceupdated 4mo agoView on Hugging Face
0likes
App README

Semantic Search — Sociology Papers

Semantic search over ~24,000 sociology paper titles and abstracts, using sentence embeddings and FAISS.

These are the papers: https://docs.google.com/spreadsheets/d/1yToQO_-fWZhKcfvkTTUGf0MciSNPEJlwMSWSuJUz1sE/edit?usp=sharing

Install

bash
pip install sentence-transformers faiss-cpu pandas numpy

Build the index (one-time)

Place sociology_papers.csv (tab-separated) in the same directory, then run:

bash
python embed.py

This produces two files:

  • —index.faiss — FAISS vector index (cosine similarity, 384 dimensions)
  • —papers.pkl — metadata for each paper (title, authors, year, journal, doi, abstract)

Takes a few minutes for ~24,000 papers.

Search

bash
python search.py "how do markets make emotions"
python search.py "race and inequality in education" --top 10

Each result shows rank, similarity score, title, authors, year, journal, DOI, and an abstract snippet.

Data

Source CSV: sociology_papers.csv (tab-separated)

ColumnRole
titleEmbedded (combined with abstract)
abstractEmbedded; shown as snippet in results
authorsShown in results
yearShown in results
journalShown in results
doiShown in results

Model

all-MiniLM-L6-v2 via sentence-transformers — fast, 384-dim embeddings, strong semantic similarity performance.