Company recognition from text - Githubissues

okiriza / esg-paper

MIT License

0 stars 0 forks source link

Company recognition from text #1

Open okiriza opened 1 year ago

okiriza commented 1 year ago

Need a function in nlp/ner.py that:

Input: given a body of text
Output: returns a list of company mentions in the text.

Each company mention consists of (start, end of the text position; company ID).

Alternatively:

Input: given a list of sentences
Output: for each sentence, a list of company mentions in the sentence.

@reinhack 23 Aug 2023 RE: pull request list company:

[ ] list companynya perlu dibuat alias/nama yg simpel/sering muncul di berita supaya bisa menangkap company mentions secara luas. contoh: "Telkom Indonesia (Persero) Tbk" --> "Telkom" (case sensitive), "Bank Pembangunan Daerah Jawa T" --> "BPD Jatim" atau "BPD Jawa Timur". also no need for "Tbk" and/or "Persero" at the end. 1 company bisa >=1 alias
[x] sepertinya nama2nya banyak yg terpotong? contoh: "BJTM,Bank Pembangunan Daerah Jawa T"

reinhack commented 1 year ago

iya 2 poin ini sebenarnya udah ditulis as comment di notebook idx-company-detect.ipynb as "room for improvement" lagi on progress nyari datanya

perlu mutliple name setuju, mungkin karena emiten ada banyak dan gak semua banyak news nya, kita prioritasin yg besar kayak emiten2 di LQ45 dulu ya
nama yg terpotong sekarang source nya dari https://www.idx.co.id/en/market-data/stocks-data/stock-list belum nemu source lain yg akurat dan gampang narik datanya