lucene-search-engine 🔍

A Lucene implementation of indexing and searching documents from the Cranfield collection, evaluated using trec_eval.

Usage

(You will need maven, java, python x)

To download and run the code:

git clone https://github.com/nating/lucene-search-engine
cd lucene-search-engine
./run.sh

You will see the trec_eval results of the Lucene application.

Output

The output of run.sh should look like this:

Metric	Geoff's	Seamus'	Difference
num_q	225.0	225.0	0.0
num_rel	218235.0	178376.0	39859.0
um_rel	1837.0	1837.0	0.0
num_rel_ret	1771.0	1714.0	57.0
ap	0.425	0.375	0.05
...	...	...	...

These outputs represent:

• The metrics from trec_eval.
• The results of the application (Geoff's).
• The results given as an example/ground truth in the TCD Computer Science module CS7IS3 (Seamus').
• The difference between the two results for that metric.

Files

• run.sh packages the maven app, indexes the cran files, performs the cran queries across the indexes, runs trec_eval, creates a results file and outputs the results of trec_eval.
• IndexFiles.java contains the code for indexing the cranfield documents.
• CustomAnalyzer.java contains the custom analyzer for analyzing document and query content.
• SearchFiles.java contains the code for running the cranfield queries and creating the results file.