Skip to main content
25 events
when toggle format what by license comment
S Feb 8, 2022 at 21:00 history bounty ended CommunityBot
S Feb 8, 2022 at 21:00 history notice removed CommunityBot
Feb 3, 2022 at 9:55 comment added Udaya Unnikrishnan I think the problem can be classified under Named Entity Recognition(NER). Please check NER using spaCy
Jan 31, 2022 at 23:27 history edited Outcast CC BY-SA 4.0
added 123 characters in body
Jan 31, 2022 at 21:27 history edited Outcast CC BY-SA 4.0
added 54 characters in body
Jan 31, 2022 at 21:10 history edited Outcast CC BY-SA 4.0
edited title
Jan 31, 2022 at 20:01 answer added Brian Spiering timeline score: 4
S Jan 31, 2022 at 19:54 history bounty started Outcast
S Jan 31, 2022 at 19:54 history notice added Outcast Canonical answer required
Jan 31, 2022 at 19:53 history edited Outcast CC BY-SA 4.0
deleted 510 characters in body
Oct 10, 2021 at 15:04 history bumped CommunityBot This question has answers that may be good or bad; the system has marked it active so that they can be reviewed.
Jun 11, 2021 at 12:01 history bumped CommunityBot This question has answers that may be good or bad; the system has marked it active so that they can be reviewed.
May 5, 2021 at 19:00 history edited Outcast CC BY-SA 4.0
added 42 characters in body
May 5, 2021 at 17:48 comment added Outcast @NikosM., I may even agree that a rule based approach may be better than an ML one but I am asking if we do want to pursue an ML approach then which should this be (even though it may be worse in the end than the rule based one).
May 5, 2021 at 14:08 comment added Nikos M. I understand that one may want to see how ML methods might work out for a problem, even out of mere curiosity. But ML is not magic nor is it suitable for all kinds of problems. That being said most invoice texts are more or less typical (like any official document) and I doubt ML can perform magic and determine which one sent which without concrete identifiers. IMO
May 5, 2021 at 12:28 history edited Outcast CC BY-SA 4.0
added 10 characters in body
May 5, 2021 at 12:27 comment added Outcast @NikosM. there is brand name etc in the invoice but to identify you have to build an application. A rule based can work but I want to test an ML one too. I think that we deviate from the topic here by a lot. My question was how to approach a problem of this kind described above in an ML WAY.
May 4, 2021 at 15:55 comment added Nikos M. Wouldnt the invoice contain some identification of the supplier (eg brand name, address, telephone, other uniquely identifiable items). I think this is an overengineering approach, maybe something simpler (like hinted above) can be way more helpful
May 4, 2021 at 9:30 comment added Outcast @Aditya OCR sounds good and I do it but after this you have to detect the supplier name (with ML) from the whole invoice text, hence this problem above.
May 4, 2021 at 9:27 comment added Aditya If the problem is to find the supplier/shop name then can't we do OCR on the receipt image? I would assume that most shops will ha e their names righ on top of a payslip. I will be nice if you break your problem at state level probably, that should reduce the classes.
May 4, 2021 at 9:24 comment added Outcast @NikosM. please see my updated post although not sure if this would change something to your answer.
May 4, 2021 at 9:23 history edited Outcast CC BY-SA 4.0
added 303 characters in body
May 3, 2021 at 9:33 comment added Nikos M. ca you provide some example of the classes and why they are so many? Maybe another approach fits your case better than explicit classification
May 2, 2021 at 20:45 answer added Abhishek Verma timeline score: 0
May 2, 2021 at 14:44 history asked Outcast CC BY-SA 4.0