Question

how can you tell what a website is about from its source code?

Answer 1

ok so i've written a crawler and i'm storing the links in a database but i don't want all links in one database. i want specific links in specific databases. so that links who's pages are about specific topics or subjects or categories of some ideas are placed in specific databases. and to do that i need to know what the page that link is from is about.

Answer 2

or even if same database, different tables, i'd still need to find a way to place them in the right places... :/

Answer 3

You want something that can 'read' the links, 'know what they're about', and store them accordingly?

Answer 4

yes :)

Answer 5

for now, i can do the reading part, "know what they're about" is what i'm in search of atm...

Answer 6

This is artificial intelligence, specifically, natural language processing. You want something that will go through a website and store it in a finite number of categories. Either that, or you can find a gigantic table of website-category associations.

Answer 7

It has to be customized, however, since the word 'category' is rather vague and you may have it defined it differently than someone else.

Answer 8

a finite number of categories is what i plan to start with, but as time goes on, users are allowed to submit categories of their interest that are not in the default list and that might be added later depending on how many people really would need such a category

Answer 9

This is exactly what's going on at math.stackexchange.com, except that the categorization is performed by users, not parsers. This is due to the immense difficulty of processing natural language. ;)

Answer 10

the idea is certainly to make it customizable so that users get data based on personal preferences... but i guess everything starts from somewhere, that's where i'm looking for right now :/

Answer 11

are meta tags a good idea?

Answer 12

It depends on the 'context.'

Answer 13

The biggest lesson that I learned as a CS undergrad is that the field has to be insanely well-defined for it to be feasible.

Answer 14

what would you suggest i attempt?

Answer 15

Find a Ph.D., or group thereof, carrying out research in the field of natural language processing and assist them in their pursuit of the betterment of e-mankind!

Answer 16

http://cms.dt.uh.edu/faculty/chenp/

Answer 17

Notice the guy's working on a "node-based web crawler."

Answer 18

how am i supposed to help? :s

Answer 19

assist i suppose is the right word in place of "help" :/ but still ....

Answer 20

You're now in a university. ;) Let's talk to some professors!

Answer 21

well i'm guessing it's not possible that i get in contact by any chance with the person in the link you gave me with respect to his research.... :/

Answer 22

He's just one in a million Ph.D.s carrying out research in this field.

Answer 23

i see, well it seems difficult than it looks anyway. :/ looks like i have a long way ahead of me lol

Answer 24

yo...you are thinking to much into what you are trying to progress forward on...lol...
keep it simple.. and build parts and other fixs... i have a book that you might find interest in..it's called Constructing intelligent Agents with Java.. read the book from start to finish then step back , make a plan then go with it..

Answer 25

ehhh what about metatags? they should tell you what are the links about

Answer 26

Yes, the meta information can be helpful. Also if you try to get the content of the link then do some indexing for example if you know which word repeated more that can help you to know the context of this link (document). I know it's heavy, because you need to process many links so a lot of content, but it's just an idea!

Answer 27

Thanks guys :) for now I'll be using the meta tags but in later development, i may change it to something sophisticated, i just want it working okay for now :) thanks for mentioning it again x

Answer 28

Welcome