PENCARIAN DOKUMEN TEKS TANAMAN OBAT BERDASARKAN KEMIRIPAN MANFAATNYA MENGGUNAKAN ALGORITMA WINNOWING; TEXT DOCUMENT SEARCHING ON MEDICINAL PLANTS BASED ON THEIR BENEFIT SIMILARITY USE WINNOWING ALGORITHM

MURYANTO, Azhari SN

2016 | Disertasi | FMIPA

Habits that have occurred in the community in the selection of medicinal plants for treatment, generally carried out in order to see the benefits of these plants, then paired with symptoms of pain that is felt most is most similar to the symptoms are plants that will be selected for treatment. The system is built to support a medicinal plant data search based on the symptoms, based on the principle of fingerprint similarity of these two documents. Fingerprint of document is determined using algorithms Winnowing. Winnowing algorithm uses the principle of over lapping of k - gram to obtain a hash value of the document, and then use the shift window to select a minimum hash value of each window to get a fingerprint of the document. As for determining the degree of similarity between pain symptoms document and benefits of medicinal plants document, then used a formula Jaccard Coefficient, which the number of same fingerprints are divided by the total number of fingerprints. The results of the performance tests these systems have the performance is quite good, as shown by the value of the average of precision 93.60%, the average of recall 80.60%, and the average of acuracy amounted to 97.18%.

Kata Kunci : medicinal plants, Winnowing, fingerprint, similarity, Jaccard Coefficient