Named Entity Recognition coherence

[jesus-seijas-sp]

Describe the bug
Named Entity Recognition is not working as expected:

• If an enumerated entity is added, it only finds the first occurance
• If there is a regular expression entity that overlaps with an enumerated entity, if returns both instead of finding edges

To Reproduce

const { NlpManager } = require('node-nlp');

const manager = new NlpManager({ languages: ['en'] });
manager.addRegexEntity('mail', /\b(\w[-._\w]*\w@\w[-._\w]*\w\.\w{2,3})\b/ig);
manager.addNamedEntityText('location', 'barcelona', ['en'], ['Barcelona', 'Barna']);
manager.addNamedEntityText('location', 'madrid', ['en'], ['Madrid']);

const result = manager.process('en', 'My mail is barcelona@barcelona.es and i live in madrid', {});
console.log(result);

Expected behavior
Currently it returns:

[ { start: 11,
       end: 20,
       levenshtein: 0,
       accuracy: 1,
       option: 'barcelona',
       sourceText: 'Barcelona',
       entity: 'location',
       utteranceText: 'barcelona' },
     { start: 11,
       end: 33,
       accuracy: 1,
       sourceText: 'barcelona@barcelona.es',
       utteranceText: 'barcelona@barcelona.es',
       entity: 'mail' } ]

It should return:

[ { start: 11,
       end: 33,
       accuracy: 1,
       sourceText: 'barcelona@barcelona.es',
       utteranceText: 'barcelona@barcelona.es',
       entity: 'mail' },
{ start: 48,
       end: 53,
       levenshtein: 0,
       accuracy: 1,
       option: 'madrid',
       sourceText: 'Madrid',
       entity: 'location',
       utteranceText: 'madrid' }
   ]

[jesus-seijas-sp]

Solved by two strategies:

1. Refactored Named Entities and Manager so find several occurences per utterance
2. Implemented a reduce edges strategy: https://github.com/axa-group/nlp.js/blob/master/lib/util/similar-search.js#L242