Ahmad Choirun Najib
Institut Teknologi Sepuluh Nopember

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

Search Engine for Halal Linked Open Data Using Entity Ranking Approach Ahmad Choirun Najib; Nur Aini Rakhmawati
Proceeding of the Electrical Engineering Computer Science and Informatics Vol 7, No 2: EECSI 2020
Publisher : IAES Indonesia Section

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/eecsi.v7.2037

Abstract

Halal concept is an essential aspect of Muslim daily life. Currently, many organizations around the world provide halal certification services as known as halal certification bodies. In the majority, these organizations provide halal product information on their website. However, information is presented in different formats, such as pdf, table, and text. As a result, the user is difficult to search for information on these websites. Therefore, we develop search engine on halal linked open data to facilitate users for searching halal products. We use an entity ranking approach to retrieve relevant items based on user queries that consist of an independent-ranking and dependent-ranking method. Independent ranking employs a link-count approach to indicate the information richness of the entity. Dependent ranking employs term frequency-inverse entity frequency  (TF-IEF) to measure the similarity of an entity based on terms. We use Apache Lucene to perform indexing and search process. Also, we use the Neo4j graph database to save entity ranking computation results. The results show that the system delivers excellent results. The Mean Average Precision (MAP) for top-5 results is 91,2%.
Perbandingan Metode Lexicon-based dan SVM untuk Analisis Sentimen Berbasis Ontologi pada Kampanye Pilpres Indonesia Tahun 2019 di Twitter Ahmad Choirun Najib; Akhmad Irsyad; Ghiffari Assamar Qandi; Nur Aini Rakhmawati
Fountain of Informatics Journal Vol 4, No 2 (2019): November
Publisher : Universitas Darussalam Gontor

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21111/fij.v4i2.3573

Abstract

AbstrakPenggunaan media sosial semakin hari semakin meningkat. Salah satu media sosial yang popular saat ini adalah Twitter. Menjelang pemilihan Presiden Republik Indonesia semakin banyak tweet yang membahas tentang kegiatan tersebut. Hal ini menyebabkan topik kampanye pemilu memiliki peluang yang baik untuk dilakukan proses analisis sentimen. Saat ini, mayoritas analisis sentimen di Indonesia dilakukan hanya menilai sentimen dari kalimat tanpa mengetahui apa entitas yang ada dalam kalimat. Tujuan penelitian ini yaitu melakukan analisis sentimen dengan pendekatan berbasis ontologi. Ontologi digunakan dalam menyaring data yang akan digunakan. Ontologi dalam penelitian ini adalah ekonomi dengan atribut finansial, lapangan kerja, dan kesejahteraan. Proses analisis sentimen dilakukan dengan metode Lexicon-based dan Support Vector Machine (SVM). Proses akuisisi data diperoleh sejumlah 700.000 tweet. Koleksi tersebut diseleksi berdasarkan ontologi ekonomi menghasilkan 16.998 tweet dan dilakukan pelabelan manual sebanyak 1.600. Kemudian dilakukan pengolahan data hingga diperoleh dataset final sejumlah 1.050 tweet. Berdasarkan hasil penelitian yang dilakukan akurasi yang diperoleh berdasarkan metode Lexicon-based adalah 39% dan metode SVM sebesar 83%. Dari penelitian ini diketahui bahwa SVM mempunyai performa yang lebih baik dibandingkan dengan Lexicon-based. Hasil Lexicon-based menunjukkan bahwa sentimen pada mayoritas atribut berupa netral. Sedangkan hasil SVM menunjukkan bahwa sentimen pada mayoritas atribut (finansial dan kesejahteraan) berupa positif, sisanya (lapangan kerja) berupa netral. Selanjutnya, proses ekstraksi dan pembuatan ontologi Bahasa Indonesia secara semi-otomatis pada dataset perlu untuk dikembangkan pada penelitian berikutnya untuk menyempurnakan ontologi.Kata kunci: Analisis Sentimen, Twitter, Ontology, SVM, Lexicon Abstract[Comparison of the Lexicon-based and SVM Method for Ontology-Based Analysis of the 2019 Presidential Election Campaign on Twitter] The use of social media is increasing. One of the most popular social media is Twitter. Towards the election of the President of the Republic of Indonesia, election topic tweets discussed almost every day. Hence, it is suitable for the sentiment analysis process. Nowadays, the sentiment analysis is only evaluating the sentence without knowing what the entity is in the sentence. To overcome this drawback, we propose a sentiment analysis based on ontology. Ontology is used to filter the data to be used. The ontology used in this study is economics with attributes, i.e., financial employment, and welfare. The sentiment analysis process is carried out using the Lexicon and Support Vector Machine (SVM) based methods. The process of acquiring data obtained 700,000 tweets. The collection was selected based on economic ontology to produce 16,998 tweets, and 1,600 manual labels were labelled. Then, the number of the final dataset is 1,050 tweets. The results show that the accuracy of the Lexicon-based method is 39%, and the SVM method is 83%. The SVM has better performance than Lexicon-based. Lexicon-based results show that the sentiment on the majority attributes is neutral. While the SVM results show that the sentiment on the majority attributes (financial and welfare) is positive, the rest (employment) is neutral. A semi-automatic ontology extraction and development for Bahasa Indonesia is necessary for the future works to make a comprehensive ontology and provide better results. Keywords: Sentiment Analysis, Twitter, Ontology, SVM, Lexicon