The task is to build a search engine which will cater to the needs of a particular domain. 
IR model is fed with documents containing information about the chosen domain. 
It then processes the data and build indexes. 
Once this is done, the user can give a query as an input.
Search engine then returns top 10 relevant documents as the output. 
Dataset: A Full-Text Learning to Rank Dataset for Medical Information Retrieval provided by Springer.

About the search engine :- 
The project implements Probablistic Information Retrieval System using JAVA. For easy user - document interaction GUI has been implemented, which shows the output (relevant documents) of the query in the form of a list. The user can right-click on a list item and select ‘show’ to view the content of the document. The content of the document will be shown in a message dialog box. The entire GUI is made by using swing library in NetBeans.

Methodology:-
	•	Tokenization:-  Built-in classes namely PTBTokenizer,CoreLabelTokenFactory and CoreLabel available in edu.stanford.nlp.process package was used for tokenizing the query and the documents.
	•	Corpus:- Corpus for the assignment is defined in resources/corpus/datasets.json.A json file was retrieved after unzipping the folder available in the above link.To extract the documents from this file ,we used the JSONArray class available in org.json package.
	•	Stemming:-After Tokenization the tokens were stemmed using the Porter’s Algorithm.
	•	Its implementation was done using a class Stemmer provided by opennlp.tools.stemmer package.
	•	Metaphone:-This is a phonetic algorithm,which does a better job than Soundex algorithm
           of matching words and names which sound similar.It is implemented by method encode() of class org.apache.commons.codec.language.Metaphone which takes the stemmed tokens as its argument.
	•	Term Frequency Calculation:-All the tokenized,stemmed and metaphoned distinct words form the terms of dictionary of individual documents.The dictionary is implemented using the TreeMap data structure ,which provides an efficient means of storing key(term)/value(term frequency) pairs in sorted order and allows rapid retrieval.
	•	ArrayList Implementation:-All the TreeMaps of individual documents are listed together using data structure ArrayList.It is used to support dynamic arrays that can grow as needed.
	•	Total Number of Documents Calculation:-Using the size() method of ArrayList,total number of documents were retreived.
	•	Document Frequency Calculation:-A TreeMap is used having key(all the distinct terms present in ArrayList)/value(their document frequency) pairs.
	•	Rank Calculation:-To obtain the rank of each document ,the formula-
	•	                                                               
	•	
	•	
	•	
where, 
	•	          w: term;
	•	          D:dictionary comprising all the distinct terms;
                      Q:query taken as input by user;
                      N:total number of documents available in the corpus;
                      Nw:document frequency of the term
           To display the ranked documents in descending order (in order of relevance to the query) a    
           sortByComparator() method was defined.