py-entity-linking 0.0.9

A lightweight library for entity linking in English

Lightweight entity linking solution for the English language.

Please consider citing our works if you use code from this repository. Also, we recommend using a Colab T4 GPU for faster results.

Main dependencies

• python>=3.10
• numpy==1.26.4
• SPARQLWrapper==2.0.0
• sentence_transformers==3.1.1
• aiohttp==3.9.5
• openai==1.54.2
• beautifulsoup4==4.12.2
• fake-useragent==1.5.1
• nest_asyncio==1.5.8

Example & Usage

from linking import main

# Your API token which can be found here (https://github.com/marketplace/models/azure-openai/gpt-4o)
api_token = "YOUR_API_TOKEN"

main.EL(api_token=api_token, sentence="We used PCA and FA for our experiments.", mention="PCA and FA", single="No", combination="No", disambiguation="Yes", embedding_model="Lajavaness/bilingual-embedding-large")

The correct entity for 'PCA' is:

Wikipedia: https://en.wikipedia.org/wiki/Principal_component_analysis

Wikidata: https://www.wikidata.org/wiki/Q2873

DBpedia: http://dbpedia.org/resource/Principal_component_analysis



The correct entity for 'FA' is:

Wikipedia: https://en.wikipedia.org/wiki/Factor_analysis

Wikidata: https://www.wikidata.org/wiki/Q726474

DBpedia: http://dbpedia.org/resource/Factor_analysis

Execution Time: 00:00:35

Parameters

• api_token: Your API token from here. (Required)
• sentence: An English text. (Required)
• mention: The mention you want to perform the linking, the mention should be from inside the provided sentence. (Required)
• disambiguation: Used when the mention has acronyms or the mention has two different entities inside (e.g. PCA and FA), (deafult="Yes"), (Values: "Yes", "No"). (Optional)
• single: Usually used for difficult mentions, it searches each word of the mention individually, (deafult="No"), (Values: "Yes", "No"). (Optional)
• combination: Usually used for difficult mentions, it makes combinations for each word of the mention, (deafult="No"), (Values: "Yes", "No"). (Optional)
• embedding_model: A sentence-transformers model to perform text similarity, (deafault="Lajavaness/bilingual-embedding-large"), (Values: str of the name of any sentence-transformers model). (Optional)

Licence

This library is licensed under the CC-BY-NC 4.0 license.