cs7is3-cranfield-search

Project developed as part of CS7IS3-Information-Retrieval-and-Web-Search coursework

A search engine implementation for cranfield collection using Apache Lucene(Java) Library. The dataset is a collection of 1400 documents, and the query file consists of 225 queries.

Getting Started

Instructions to setup the project and to run the search engine

Requirements

Ubuntu ~18.04
OpenJDK 1.8.0_242
Maven 3.6.0
GCC 7.4.0

Extract the project to your home path

$ tar xzvf cs7is3-cranfield-search.tar.gz ~/
$ cd ~/cs7is3-cranfield-search

Compile

$ mvn clean package -DskipTests

Run

$ java -server -Xmx2g -cp ./target/cranfield-1.0.jar ie.tcd.cs7is3.cranfield.App

The default analyzer used is the EnglishAnalyzer and the similarity used is BM25Similarity. The data document path, query path, analyzer and similarity can be changed by the -d, -q, -a, -s command line options respectively. Run the program with the -h flag for more details.

The output of the execution is stored in ./output/results.txt

Evaluate

To score the search engine trec-eval is used.

$ cd trec_eval-9.0.7
$ make
$ ./trec_eval ./data/QRelsCorrectedforTRECeval ./output/results.txt 2>&1 | tee ./output/trec_eval_score.txt

Or

Run the execute.sh to do all three in one go.

$ sh ~/cs7is3-cranfield-search/execute.sh

Menotron/cs7is3-cranfield-search