entitizer / concepts-parser-js

Nodejs module for Extracting Concepts from text.

concepts-parser

Nodejs module for extracting concepts from text.

A Concept is a part of a text that may be a Named entity. We use Concepts for learning new named-entities, for searching known entities, for identifying entity names(synonyms, abbreviations), etc.

Usage

JavaScript:

const parser = require('concepts-parser');

const concepts = parser.parse({ text: 'Some text', lang: 'ru', country: 'ru' });

TypeScript:

import { parse } from 'concepts-parser';

const concepts = parse({ text: 'Some text', lang: 'ru', country: 'ru '});

API

parse(context, options)

Finds concepts in a context.

• context (Object) required - Context
  • text (String) required - Text to find concepts;
  • lang (String) required - Text language, 2 chars code: en, ru;
  • country (String) optional - Context country: ru, it;
• options (Object) optional:
  • mode (String) optional - Can be identity or collect. Default: identity. identity mode excludes filters: start_word, duplicate and partial;
  • filters (String[]) optional - Ordered list of filters;

Valid filters

1. invalid_prefix - deletes invalid prefixes;
2. invalid - exclude invalid concepts;
3. partial - exclude partial concepts;
4. prefix - add prefixes to concepts;
5. suffix - add suffixes to concepts;
6. start_word - exclude sentence start words;
7. abbr - finds concepts abbreviations;
8. known - finds known concepts;
9. duplicate - exclude duplicates;
10. quote - concats concepts in quotes: Teatrul Național "Mihai Eminescu";