Text Similarity Detection Between Documents Using Case Based Reasoning Method with Cosine Similarity Measure (Case Study SIMNG LPPM Universitas Sriwijaya)

Abstract

LPPM Universitas Sriwijaya is an institution that coordinates academic research and community service inside Universitas Sriwijaya. In carrying out the duty, LPPM assesses every proposal’s originality which would be impossible to do manually in the future due to massive data growth. Thus, automatization for the proposal's originality check is needed. The Case Based Reasoning method is used in this research because it allows the system to reuse the information that has been obtained to find documents that are similar to the test document. In this study, the data is represented in the form of the Vector Space Model and uses Cosine Similarity to measure document to document similarity. The data is represented by giving weight for each part of the tested documents. In this study, four formulas from previous research will be used for term weighting then the final result will be compared. The process begins by extracting data, separating parts of the document, figuring the similarity value of the test document to the case base utilizing Cosine Similarity Measure, results filtering with a certain threshold, summarizing the calculation results, and finally preserving the results obtained to be reused in the next calculation. The results of this study indicate that the text-similarity detection between documents has been successfully carried out using the proposed method with the best sensitivity level and the fastest computation time achieved in configuration II.