Information Retrieval System
ATTENTION!!!
The whole work IS NOT done by myself. This is a course project. All the codes are provided by my teacher. And maybe the original authors are T.S. Chua and Shen Mo Jun.
All rights reserved by the authors. Should this violate your rights, kindly remind me. I shall delete your document without hesitation and delay.
This is the original usage of the project
INSTRUCTIONS TO USE THE PROGRAMS
================================
To uncompress the program:
tar -xvf irs3.tar
There will be some files in the main directory and two sub-directories named "data" and "indexdb" respectively.
There are 5 sample documents in the directory of "data".
To compile the program:
- Go to the your main directory where the programs are stored and run:
make
- to make irmain
and make index
to make irindex
To execute the program:
- Go to sub-directory "indexdb".
Modify the first line of "filelist.dat" to include the full path to the directory where your list of data files are stored
- Go to sub-directory "indexdb"
run ../irindex
to create the indices: this will take a while when computing weights for vector space model.
run ../irmain
to try out different queries.
For Linux, use the version under irslinux.tar:
It is a little bit different from the Unix version. It works on Linux Mandrake which uses gcc-2.95.2 and g++-2.95.2.
For other linux versions, the location of "ndbm.h" may change. In Mandrake we use "#include<db1/ndbm.h>" (in "filesys1.h"). You may probably (maybe, not sure) need to change two lines in Makefile for generating irmain and irindex. In Solaris, the ndbm files are stored as .dir and .pag. But in Mandrake, the files are stored only as .db. After changing the path of including the ndbm.h, the compiling procedure should be similar as mentioned above.