Jurnal Nasional
| Judul | Pembangunan sistem temu balik informasi (information retrieval) dalam pemilihan pemain sepak bola berkualitas di indonesia berbasis analisis sentimen. |
| Penulis: | Wina Witanti, Herry Rahmanto, Faiza Renaldi. |
| Journal | Seminar Nasional Teknologi Informasi dan Komunikasi 2016 (SENTIKA 2016) |
| Masalah Penelitian | Sepak bola merupakan salah satu olahraga yang sangat disukai dibandingkan olahraga lainnya.Sepak bola mengutamakan strategi dan kerjasama tim.Di perlukan pemain yang berkualitas sehingga memerlukan sistem yang mampu merekomendasikan 11 pemain yang sepak bola terbaik dan data yang digunakan berjumlah 120 artikel serta 70 pemain sepak bola. Pada penelitian ini menggunakan metode Naïve Bayes Classifier dengan beberapa tahapan yaitu Convert Emoticon, Cleansing,Case Folding, Convert Negation, Tokenizing, Stemming. Adapun hasil uji coba menunjukkan bahwa perangkat lunak memiliki penilaian fungsi-fungsi yang ada pada sistem ini yaitu 89,26% sistem dapat berjalan. |
| Tujuan Penelitian | Menghasilkan sistem yang mampu merekomendasikan 11 pemain sepak bola terbaik. |
| Variabel Penelitian | 50 pemain sepak bola dan 80 artikel tentang sepak bola dari situs Goal.com |
| Definisi Operasional dan Pengukuran | Information Retrieval adalah sebuah metode untuk mengambil data terstruktur yang tersimpan dalam koleksi dokumen, kemudian menyediakan informasi yang diperlukan. Tujuan dari sistem Information Retrieval adalah untuk mengambil dan menampilkan dokumen yang relevan dengan pengguna (query). Naive Bayes Classifier merupakan salah satu metode machine learning yang menggunakan perhitungan probabilitas. |
| Metodologi Penelitian | Tahapan penelitian tersebut antara lain yaitu observasi angsung terhadap pemain, memantau dari rekaman pertandingan,membaca informasi pemain dari media, melihat data statisitik dan sejarah pemain. |
| Model Penelitian | -Melakukan analisis sistem bertujuan untuk mengetahui kategori pemain sepak bola bekualitas di indonesia.Pertama,pelatih mengumpulkan data pemain sepak bola dari berabgai klub untuk proses penyeleksian. Setelah itu staff kepelatihan mencatat data pemain sepak bola dari usulan pelatih. Setelah mencatat data pemain sepak bola lalu staff kepelatihan membuat catatan dokumen berupa laporan daftar pemain sepak bola serta melakukan proses seleksi terhadap pemain sepak bola.
-Perancangan sistem yang berperan di sistem yaitu Staff Kepelatihan, untuk dapat memperoleh informasi tersebut dibutuhkan data pemain sepak bola antara lain berupa nama pemain dan posisi. Hal tersebut bertujuan agar memudahkan pelatih dalam melakukan pemilihan pemain sepak bola berkualitas di Indonesia. -Kemudian pengujian Dataset dari data-data yang sudah di dapatkan. |
| Hipotesis (Jika ada) | – |
| Hasil Penelitian | Dari perhitungan dataset dapat disimpulkan bahwa kategori dari teks uji 1 yaitu termasuk kategori positif, karena nilai probabilitas teks uji 1 pada kategori positif (3,184*10-9) lebih besar dari nilai probabilitas teks pada kategori negatif(1,026*10-8). |
| Implikasi Manajerial | Penelitian ini menghasilkan sistem yang dapat merekomendasikan 11 pemain sepak bola berkualitas dan terbaik berupa pola strategi dalam formasi. Dalam penelitian ini digunakan 2 kriteria yaitu nama pemain dan posisi. |
| Keterbatasan Penelitian | Jumlah data yang diolah diharapkan banyak dari sebelumnya,agar memberikan formasi pendukung menangani masalah dalam pencarian terhadap pemain sepak bola.Proses hasil pencarian pemilihan pemain dan penentuan lin up formasi membutuhkan waktu yang cukup lama tergantung banyaknya data. Pada kategori klasifikasi kata diharapkan memiliki unsur kata yang sering muncul di dalam suatu artikel atau berita, sehingga proses yang dapat dilakukan dengan menggunakan metode lain bisa menghasilkan persentase angka yang maksimal. |
| Agenda Penelitian yang akan datang | – |
Jurnal Internasional
Journal Title: A review on the application of evolutionary computation to information retrieval
Author: O.Cordon, E. Herrera-Viedma, C. Lopez-Pujalte, M. Luque, C. Zarco
Journal: International Journal of Approximate Reasoning 34 (2003) 241-264
Research problem:
Apply Artificial Intelligence (AI) technique to IR to solve commercial IRS problems. Commercial IRs, usually based on the Boolean IR model, have given unsatisfactory results. Space vectors, probabilistic and fuzzy models, which have been developed to expand the Boolean model, as well as the application of knowledge-based techniques, have solved some of these problems, but there are still some shortcomings.
Research purposes:
1. Analyze the various types of IR problems that have been solved by EA
2. Describe some specific approaches proposed and evaluate the results obtained.
Research variables: evolutionary computing for information retrieval.
Research methodology:
This research was conducted by analyzing several approaches related to IR problem solved by EA.
Research result:
• Application of Evousionary Algorithm for information retrieval
EAs do not specifically study algorithms but they offer independent and powerful searching skills that can be used in many learning tasks, since learning and self-organization can be considered an optimization problem in most cases.
EAs has been implemented to address the following IR problems:
(1) automatic document indexing,
(2) grouping of documents and terms,
(3) query definition,
(4) learning matching function,
(5) image capture,
(6) design user profiles for IR on the Internet,
(7) classification of web pages,
(8) agent design for internet search.
• Automatic document indexing
Some approaches to automated document indexing include:
A. Gordon’s approach
B. Vrajitoru’s approach
C. Approach Fan, Gordon and Pathak
• Clustering documents and requirements
There are two different approaches in document clustering:
A. Robertson and Willet’s approach
B. Approach gordon
• Query definition
Each approach in IR uses EA such as relevance feedback techniques or like the Inductive Query by Example (IQBE) algorithm.
IQBE is proposed as ” a process in which searchers provide examples of documents (examples) and algorithms induce (or learn) key concepts to find other relevant documents ”. In this way, IQBE is a process to assist users in the process of formulating queries that are done by machine learning methods. It works by taking a relevant (and optionally, irrelevant) set of documents provided by the user and applying the off-line learning process to generate queries automatically that explain the needs of the user.
• Matching function learning
Two different variants have been proposed in the particular literature:
A. A linear combination of similarity functionality exists. The algorithm is tested on the Cranfield collection and the results look very encouraging.
B. Automatic equality measurement of learning (with GP algorithm). No experiments were shown.
• Image retrieval
There are several proposals that combine EAs and shooting, among others:
A. Cho and Lee developed an image-capture system based on human preferences and emotions using an interactive genetic algorithm (IGA) with the aim of complementing the lack of user expression capabilities. The system extracts features from images with wavelet transforms, and uses IGA to search for user-owned images by using the user’s choice as fitness when the fitness function can not be defined explicitly.
B. Kato and Iisaku explained the shooting system is also based on GA with an interactive mechanism that dynamically reflects the subjectivity of individual users in search results and proposes a new method for increasing indirect dependence.
C. The local pattern of similarity is based on the idea that distinguishing different objects in images requires different sameness criteria for each object. Additionally, GA-based methods are proposed to find the assignment of optimal equality criteria to the image area, and incorporated in the feedback mechanism of relevance.
• Design of user profiles for information retrieval on the Internet
Three proposals involving user profiles and GA, among others:
a. Agents are proposed to model the user’s information needs for search on the web through GA-based adaptation processes with fuzzy genes.
B. Scheme to maintain experience-based knowledge about user preferences in user profiles.
C. An adaptive software query system (ASQ) to improve the accuracy of user profiles.
Classification of web pages
Loia and Luengo present an evolutionary approach that is useful for cataloging automatically and also for classifying web documents.
• Design agent for internet search
A. Bergstr oomom et al. Presents a method for finding textual relationships in electronic documents using GP and semantic networks. This system aims to protect IR by automatically extracting relationships from text. Experiments were performed on pre-processed text from the web and preliminary results confirm the feasibility of the method.
B. Chen et al. Implementing a private Internet spider based on the best first search and GA techniques.
C. Walker uses GP to search the web. In addition, it also uses parallel implementations. The results show that databases for different search engines are stable upon completion of initial searches, ie the user can optimize the initial results by repeating the search over a given period of time.
Limitations of the study:
The study only reviewed some of the applications of evolutionary computing for IR Bottom of Form
file Review:
link Download Jurnal
Jurnal Nasional : PEMBANGUNAN SISTEM TEMU BALIK INFORMASI (INFORMATION RETRIEVAL) DALAM PEMILIHAN PEMAIN SEPAK BOLA BERKUALITAS DI INDONESIA BERBASIS ANALISIS SENTIMEN
Jurnal Internasional : A review on the application of evolutionary computation to information retrieval