Information Retrieval System for Searching JSON Files with Vector Space Model Method

摘要

The purpose of this research is to build a retrieval system for searching files or documents in JSON or JavaScript Notation format. The source of JSON files that searched comes from the generated journal files that have been extracted from the title and the keywords. This retrieval system is designed using the Vector Space Model method to obtain optimal searching results by looking for similarity of query words with the documents sought. This system is made with Java programming based object-oriented so that it can be implemented on many platforms and operating systems. This system reads all the words in the JSON file, then filters it for the stopword list to remove it. Then the stemming process is carried out to find the basic word. Stemming results are indexed to store the number of word frequencies in each document, the indexing results are stored in memory to speed up the searching. When a user searches, the system will search for indexing results based on query words, then weighting TF-IDF each word searched and calculated similarity with the Cosine Similarity formula with related documents. All related documents will be sorted descending based on the scoring value of the similarity. The word search process in this system more or less takes 150-300 milliseconds in JSON documents which amount to 140.