Hii ,I am Shrilakshmi from India .I am interested in participating for GSOC 2020.
I am intended to work on 7 th topic i.e Search, under that on fuzzy search.
For non-english speaking people in India, it is hard to translate the search the need in modern days of internet world. It is important for the search engines understanding the non-English language content. There are translator software which would translate the few of the language to English, having the content translated to English is easy to search on English content catalog. But correcting the word which is wrongly interrupted translated words, will give tremendous capablity
My idea is to build a neutral network model to autocorrect indian language which can be extended to other languages.
Its bit late as I was doing home work on the topic.
Hi and welcome.
thank you for your idea that goes in the same direction I came up with in the search topic, really looking forward for more details of your idea.
I suggest you read it up and detail your idea
Thanks for the replay.For sure I will go through your idea.
So far I have built a neural network for kannada lang spell checking (hoping to extend it ) using tensor flow.It works like ,whenever user enters kannada word written in english will be matched to the exact dictionary word which has phonetic relation.
The project what I have built so far is as follows :
Topic:‘Neutral network model to autocorrect indian language’
It takes two input :
1.Given a word with language will be matched to the correct word
Given a record or document, word which is misspelled or doesn’t match to the context will be corrected.
So far I have worked on first input form and got 83% accurecy for about 3000 words (trying to get more words) .Now I am preparing for second input form(NLP).
Hoping to get opportunity to learn and contribute in this summer.
Thanks for the time and I wish you the best.
Thanks for the suggestion.I will do it very soon .I have two queries:
1.I didn’t get what is code base.Is it raising pull request in git.If so I was puzzled to raise an issue as joplin is in js (I have written my code in python ) and everyone discuss there some improvements along with screen shots how it works ,but I have built API kind where user needs to enter some word and will be matched to exact word that too for Kannada .
Can you please help me regarding how to do it as its my first time to work with commuinty (bit scared).
2.If its not git pull request may I know what it is??
Thanks for the time and I wish you the best