Measuring Meaning Similarity Using TF/IDF and Term Synonym ID
Ardi Sanjaya, Ahmad Bagus Setiawan, Danar Putra Pamungkas, Intan Nur Farida, Made Ayu Dusea Widyadara · 2023
This work proposes a novel alternative for measuring the meaning similarity between two sentences in Bahasa Indonesia. Two sentences are considered to have the same meaning if each word in those sentences has the same synonym. In this regard, a synonym database is needed. In this study, the synonym database was developed by categorizing words with the same meaning. Each word group was given a unique ID. The first scenario was performed without stemming, while the second was performed with stemming during the pre-processing. Each word in the sentence was matched to the synonym database using the synonym ID reference. The synonym ID was weighed using TF/IDF, and the vector distance was measured using Cosine similarity. The most optimal test result was noticed in the first scenario with a 89.72% similarity score out of 80 data tests, higher than the second scenario with an 76.24% similarity score.