Looking for some suggestions.

[Palestinian]

Hello everyone,

I'm thinking about starting a new project that will be used to 'import' a huge amount of data (patient cases) from a certain website to be used for searching porposes.

Each case got a unique case ID added to the URL, example:

http://raisunhcr.org/user/Member_Biodata.aspx?caseid=237602

 

I've been trying to contact the UNHCR team in hope of getting the database from them but their number is busy atm.

What is my best approach for such a project? What function might do the job better than other functions? is it recommended to store the data in excel sheets and work with them or is there a better/faster way to accomplish that task? ' I considered SQLite but it sounded way too gibbrish for me. '

At the moment there are 326621 cases that I need to import, they increase by -+300 on daily bases, I will also be needing to check if any data changed in the files that I already imported by comparing them with the website version of the files (assistance gets added to cases every day)

[JohnOne]

I'd leave the database in whatever format it already is.

[Palestinian]

I just got off the phone with them, they don't mind the idea but they won't give me the database for security reasons ( I already got access to the information :/ ) so now I will be creating my own database using whichever format I see fit.

[JohnOne]

Can you not just query the database online?

[Palestinian]

I can, however a normal search takes some time depending on the server load and the internet connection and the accuracy of the searching terms, the reason I'm doing this is to have an offline searching engine that only requires an internet connection once to compare the data then work offline the rest of the day, the data doesn't even need to be updated daily, it works if the data is updated once a week too.

[JohnOne]

 I'd use SQL of some flavour then, lite, MY etc...

Probably take you a week to update one third of a million records from individual web pages whatever format you use.

[Palestinian]

I will try and read about SQLite again, hopefully I'll have a better mindset for it, I know it might take some time to fully update the records, but as I said a weekly update works just fine as well, thank you.

[orbs]

if searching is your goal, then possibly store the text of each webpage into a text file, and have it indexed by the operating system. if you deal with a database with O(10^6) textual records, each can be presented in a single webpage - that shouldn't take that much space.

[Palestinian]

Simply storing the text of each webpage won't work, search doesn't return a unique page if the case size is bigger than 1 (case size is determind by family members number), only way to get a unique result is by searching for Individual ID's and then using the export button, however that only exports the main information (name, ID number, DoB ), it doesn't import the actual needed data (health assistance, case assistant, medical coverage, payments, etc.)

The problem with the search function on the website is that it sends you to the PA's case page, not the actual case that you are looking for, then you'll have to select the name of the patient you are looking for from the family's dropdown menu which in return updates the information on the page.

[jchd]

SQLite is most probably the best tool for your use case, much, much preferable to Excel.

Can you dump the database, or at least capture rows or tables? Anyway, once you have a workable input way like a .CSV or something like html tables, it will be pretty easy to design a database schema to host your data efficiently. Details will be needed to make a more precise advice/guidance but in all cases there is a solution waiting.

Edit: to be fair, I'm known to be an almost unconditional SQLite fan (with robust reasons).

Edited April 24, 2014 by jchd