Recently active topics in LUCI

If you've found this website then congratulations. This is the class forum and place for feedback on the class. Ask questions here, discuss related topics, whatever is appropriate in conjunction with the University of California, Irvine's Fall 2008 course on Information Retrieval.

There's is a neat little article on a risk-like game in the NY Times today. To summarize, it's a risk-like game that takes place in the real world portraying the sense of a ubicomp atmosphere at universities and workplaces around the country. Link is below:

http://www.nytimes.com/2008/03/21/tec...

This is the forum that we are going to use for discussions about the class.

Here are the components you will probably need to build. I say "probably" and "should" because you can do it however you want. The evaluation doesn't specify an architecture and oftentimes other people have good ways of doing things that I don't consider....

1)
Something that can read in a bunch of files (either from a list or by looking in a directory).
I would start by seeing if java.io.File can help there.

2)
Then you need to read in each of the files line by line and parse out the content. Depending on the content you will need to do different things. If you are using the sample output files, then for every document URL you read in you will have to update the data structure with what you find. The posting list is just a data structure put together the correct way.

You will need to use the java.util.Map interface which maps from a document URL to a term frequency in it. (The number of times a term appears in that document) Alternatively you could use a list. (I don't recall which Jam recommended in discussion) Call this the "DocTFMap".

Then you will need a java.util.Map data structure which maps from a term (stored as a string) to the DocTFMap's. Call this the TermMap

When you choose an implementation for the Map data structures, you should probably choose a map that keeps things alphabetized for efficiency.

You will process a document at a time.
For each document, count the number of times each term occurs in it. (term frequency)
When you figure that out, find the right DocTFMap in the TermMap by using the term you are currently processing. Then insert a new Document TermFrequency pair into it.

3)
You should write out your data structure to disk using Serialize utilities.

4)
You should write a second program which reads in the data structure asks for the user for a term and then returns the list of documents that have that term. (Again a lookup into the TermMap)

5)
All those components again for the Image lookup dictionary, but just one Map is necessary which maps from Document URLS to Images. Much of the code will be very similar.