KLASIFIKASI MULTI-LABEL TEKS HADIST DENGAN MENGGUNAKAN K-NEAREST NEIGHBOR DAN DIINTEGRASIKAN LABEL POWERSET

SAID, IAN APRILIANTO (2023) KLASIFIKASI MULTI-LABEL TEKS HADIST DENGAN MENGGUNAKAN K-NEAREST NEIGHBOR DAN DIINTEGRASIKAN LABEL POWERSET. Undergraduate Thesis thesis, Institut Teknologi Telkom Purwokerto.

	Text 1. Cover.pdf Download (1MB)
	Text 3. Abstract.pdf Download (334kB)
	Text 2. Abstrak.pdf Download (332kB)
	Text 4. BAB I.pdf Download (496kB)
	Text 5. BAB II.pdf Download (1MB)
	Text 6. BAB III.pdf Download (703kB)
	Text 7. BAB IV.pdf Restricted to Registered users only Download (921kB) | Request a copy
	Text 8. BAB V.pdf Download (397kB)
	Text 10. Lampiran.pdf Restricted to Registered users only Download (741kB) | Request a copy

Abstract

Hadith is one of the foundations of Islam besides the Qur'an, Hadith is also the meaning of all the attitudes, words and deeds of the Prophet Muhammad in teaching Islam. Hadiths have various meanings, Hadiths which contain advice, information, and prohibitions that are intended for all Muslims. In the hadith there are several hadith texts that contain 3 meanings in 1 hadith, or it can also be called multi-label. With a very large number of hadith texts, at this time it really takes time to label each one hadith text, therefore a classification model is created that can group hadith texts with multiple labels. This study uses a dataset in the form of hadith which has 1064 texts that have been labeled accordingly. In the previous study, classification was carried out using the same dataset in the form of authentic bukhari hadith text with an f1-score of 0.8539 by testing using the k-nearest neighbor with binary relevance and a hamming loss result of 0.178403 by testing using a convolutional neural network with a powerset label. Basically KNN is an algorithm that focuses on pattern recognition cases, but it can also be used to carry out single-label classification, therefore in this study the powerset label method is used to classify multi-labels in the KNN algorithm in carrying out multi-label classification of the hadith texts of sahih bukhari. In this test, the best hamming loss results were obtained from using the powerset label method with a value of k = 4 and an average value of hamming loss 0.0795, and the results of using the binary relevance method with a value of k = 5 and an average value of hamming loss 0.0821, so that it can be concluded that the use of the binary relevance method is not better than the use of powerset labels in this study. Keyword: Hadits, KNN, Label Powerset, Binary Relevance, Multi-label Classification

Actions (login required)

View Item