Archive of Hong Kong Legco Votes

[siuying]

any idea on how to do this?

[alanho]

i afraid i overcomplicated this. the first question would be how to convert old record into machine readable one. either we do OCR/image processing.. and for future votes, i'm in touch some contacts that have access to LegCo Secretariat to see if they can improve the system to store machine readable format instead of pdf scannings

[sammyfung]

It is difficult to use OCR to process those scanned records into machine readable one. For LegCo meeting minutes, I checked few days ago and found that all scanned records (minutes only) from 1985 are replaced, now most of them are in 'machine readable' PDFs.

And Charles Mok said on Saturday in HKLUG/ITFest seminar that LegCo Secretariat is preparing machine readable copies of voting results, but we dunno when will be completed.

[siuying]

good news! 👍

[alanho]

any links to those machine readable PDFs?

it doesn't need to be full or perfect OCR, might be a simple histogram will do the job? it's either yes, no or abstain.

but if somebody will let us know LegCo will convert old data to machine readable one, we could just wait for a bit. if not, maybe someone could get in touch with a computer vision lecturer/professor in local U and get this task as an assignment for students ;D

[siuying]

@alanho http://www.legco.gov.hk/yr12-13/chinese/counmtg/motion/mot_1213.htm#toptbl it will need some smart algorithm to extract them into structured data, though.

[alanho]

so far.. this is all i have time for.. a script to extract all voting result into images or PDFs..

https://gist.github.com/alanho/5433521

[siuying]

not very usful if we just got the image and need manual intervention. Perhaps just write a scraper to extract data from the app: https://itunes.apple.com/hk/app/yi-yuan-biao-xian-lu/id549783193?mt=8 where they are manually input data

[alanho]

interesting app. didn't know about this app! but i guess they don't have all the voting result do they??

[alanho]

the target is to build a database that these guys can use, so they don't have to worry about data input, just focus on presenting the data in their own ways

[siuying]

I think they have all voting result in the period, but only 2008－2012