DATA MINING - EXTRACTION - SCRAPE: 6000 pages of pdf across 200 or so pdf files
Budget: $250 – $750 USD
DETAILED REVIEW FOR DATA EXTRACTION/MINING FROM PDF
I have financial records in PDF form. There are something like 150-200 different pdfs. THE pdfs are for the same form (called the 460 form) – but the information (data) in each form is, of course, distinct.
The length of the pdf file varies. The length depends on the activity during the prior reporting period. Most of the filings are between 50-100 pages; a few are 5-10 pages, while a few are 200 pages or so. Of those pages, I only need the contributions – and those are contained under SCHEDULE A (so first three pages can be deleted; and after the end of exhibit A, the subsequent pages can be deleted). SCHEDULE A typically takes up about 80-90% of the total pages.
The header on each page of the financial record has information I need to track. The info is: The entity’s name and assigned accounting number, the date of the report, and the individual associated with the report. This information is known not only from the pdf forms, but also, because each pdf is stored in a folder. That folder contains a tracking number sequence (1-48), where each number is an individual candidate and the entities number (so, eg., “09 - Jones - 1347592”)
Within exhibit A, there are five entries per page. Each entry provides the data that needs to be tracked.
In each row, the data must be sorted into these columns:
Date of donation (sometimes there are two dates, if same date, only record one; when two different dates, record only the early date);
First Name (first name of individual; if not an individual the entire entity name will go in this field)
Middle Name
Last Name
Number of address
Street name
Unit number
City
State
Zip code
Type of person/entity donating (five choices, something like IND (individual, OTH (other), COM (committee)
Occupation (only if an individual, if a “PAC” or “OTH” then there is nothing written here)
Employer (only if an individual, if a “PAC” or “OTH” then there is nothing written here)
Donation amount
Cumulative donation amount
In all, there are about 6,000 pages of financial records, with about 5 records per page. I expect 30,000 records total to be produced.
If this round goes well, there may be another round that may need to be presented too. I’m fine with trying a sample set (say 6,000 rows) first to see how it goes – rather than betting it all. I’m happy to discuss it further as necessary.
As a condition of accepting this job/contract you must agree to hold the information strictly confidential adn to delete all information once the job is completed.
On the one hand, if you have no or little experience, you should be bidding at a discount; on the other hand, if you have a long established record of completing work on time and on budget, I expect your work, due to your reputation, will cost more -- I'm fine with that. That's the market at work. BTW, I have a strong background as a clien t- but it's buried at u wrk -- and they've gotten so difficult on the client side, I'm trying other options.
I have financial records in PDF form. There are something like 150-200 different pdfs. THE pdfs are for the same form (called the 460 form) – but the information (data) in each form is, of course, distinct.
The length of the pdf file varies. The length depends on the activity during the prior reporting period. Most of the filings are between 50-100 pages; a few are 5-10 pages, while a few are 200 pages or so. Of those pages, I only need the contributions – and those are contained under SCHEDULE A (so first three pages can be deleted; and after the end of exhibit A, the subsequent pages can be deleted). SCHEDULE A typically takes up about 80-90% of the total pages.
The header on each page of the financial record has information I need to track. The info is: The entity’s name and assigned accounting number, the date of the report, and the individual associated with the report. This information is known not only from the pdf forms, but also, because each pdf is stored in a folder. That folder contains a tracking number sequence (1-48), where each number is an individual candidate and the entities number (so, eg., “09 - Jones - 1347592”)
Within exhibit A, there are five entries per page. Each entry provides the data that needs to be tracked.
In each row, the data must be sorted into these columns:
Date of donation (sometimes there are two dates, if same date, only record one; when two different dates, record only the early date);
First Name (first name of individual; if not an individual the entire entity name will go in this field)
Middle Name
Last Name
Number of address
Street name
Unit number
City
State
Zip code
Type of person/entity donating (five choices, something like IND (individual, OTH (other), COM (committee)
Occupation (only if an individual, if a “PAC” or “OTH” then there is nothing written here)
Employer (only if an individual, if a “PAC” or “OTH” then there is nothing written here)
Donation amount
Cumulative donation amount
In all, there are about 6,000 pages of financial records, with about 5 records per page. I expect 30,000 records total to be produced.
If this round goes well, there may be another round that may need to be presented too. I’m fine with trying a sample set (say 6,000 rows) first to see how it goes – rather than betting it all. I’m happy to discuss it further as necessary.
As a condition of accepting this job/contract you must agree to hold the information strictly confidential adn to delete all information once the job is completed.
On the one hand, if you have no or little experience, you should be bidding at a discount; on the other hand, if you have a long established record of completing work on time and on budget, I expect your work, due to your reputation, will cost more -- I'm fine with that. That's the market at work. BTW, I have a strong background as a clien t- but it's buried at u wrk -- and they've gotten so difficult on the client side, I'm trying other options.