TaBERT: A new model for understanding queries over tabular data • Hacker News Board

Re: TaBERT: A new model for understanding queries over tabular data

#21

Post by nl » Sat, Jul 04, 2020, 4:31 AM UTC

SheinhardtWigCo wrote: post #19

Honest version: > Why it matters: > Improving NLP allows us to create better, more seamless human-to-machine interactions for tasks ranging from identifying dissidents to querying for desperate laid-off software engineers. TaBERT enables business development executives to improve their accuracy in answering questions like “Which hot app should we buy next?” and “Which politicians will take our bribes?” where the answ…

Why are you making fake quotes here?

Most of the data in the world is in tables, and most people don't speak SQL. A very large purpose of computers has been for managing and querying this data and there is nothing nefarious at all about that.

Translating between the two is something that there have been many attempts at.

Re: TaBERT: A new model for understanding queries over tabular data

#22

Post by nl » Sat, Jul 04, 2020, 4:45 AM UTC

abhgh wrote: post #4

Does anyone know how it relates/compares to Google's TaPaS? [1] I notice this paper doesn't refer to it. [1] https://ai.googleblog.com/2020/04/using-neural-networks-to-f...

They both report performance on WikiTableQuestions (called WikiTQ in the TAPAS paper).

Wang, et.al. (previous SOTA): 44.5

TAPAS: 48.8

TaBERT: 52.3

Re: TaBERT: A new model for understanding queries over tabular data

#23

Post by louisstow » Sat, Jul 04, 2020, 11:53 AM UTC

Does anything like this exist for XML documents? Wonder if it could be used for identifying interesting information in web pages.

Re: TaBERT: A new model for understanding queries over tabular data

#24

Post by sriku » Wed, Jul 15, 2020, 5:23 AM UTC

TABERT no longer on the Spider leaderboard? - https://yale-lily.github.io/spider . The top is "RATSQL v2 + BERT" testing at 65.6 for exact matches.

Re: TaBERT: A new model for understanding queries over tabular data

#25

Post by kfk » Wed, Jul 15, 2020, 9:02 AM UTC

iandanforth wrote: post #20

ianhorn wrote: post #6

I'd like to see something that could do this, handling the awfulness of real world tabular data. "What country has the highest GDP? Okay, which table has GDP? Is it the country_gdp table? No, that's an old one that hasn't been written to in 3 years. Ah here it is, but you need to join against `geopolitics`, but first dedup the crimea data, since it's showing up in two places, we're can't remember why it got written t…

Out of curiosity have you tried any of the data catalogs? Alation, Informatica, Collibra? Not that they solve this problem for you but that make it solvable.

What do these do? I am constantly told by IT we need data catalogues, they talk about it, they invest money, but I have never seen the results. Me and my team keep most metadata info of tables in json files, not pretty but at least I can see what it does

Re: TaBERT: A new model for understanding queries over tabular data

#26

Post by myth_drannon » Wed, Jul 15, 2020, 1:06 PM UTC

IfOnlyYouKnew wrote: post #15

jhj wrote: post #3

Earlier quoted context omitted.

it's in the paper https://github.com/facebookresearch/tabert

No pretrained models, though, unfortunately.

They do provide it. The link is in github repo. Google Drive shared folder : https://drive.google.com/drive/folders/1fDW9rLssgDAv19OMcFGg...