Summarization of Speech to Text from Reporter in Police Office with Latent Semantic Analysis (LSA) Method
Accepting complaint report was a police service to the community. The writing of reported incident in report construction process was time-consuming. This made long queue emerged. One of the solutions developed was by summarizing voice of the complaint uttered by the reporter. The used summarization method was Latent Semantic Analysis (LSA) that was implemented by using Singular Value Decomposition (SVD) algorithm. In addition, the usage of cross method extraction sentence and Speech to Text from Google Speech API was expected to increase the voice to text conversion accuracy. The testing resulted in f-measure and recall values of 92.3% on the first document, while the biggest precision value was 100% with total of 10 data test. The test was performed by compression rate of 30%. The LSA method used was able to summarize by taking attention to words and sentences similarities. Nevertheless, the summarization result depended on the sentence length.