Joan Santoso

Large Scale Text Classification Using Map Reduce and Naive Bayes Algorithm for Domain Specified Ontology Building

Internet that covers a large information gives an opportunity to obtain knowledge from it. Internet contains large unstructured and unorganized data such as text, video, and image. Problems arise on how to organize large amount of data and obtain a useful information from it. This information can be used as knowledge in the intelligent computer system. Ontology as one of knowledge representation covers a large area topic. For constructing domain specified ontology, we use large text dataset on Internet and organize it into specified domain before ontology building process is done. We try to implement naive bayes text classifier using map reduce programming model in our research for organizing our large text dataset. In this experiment, we use animal and plant domain article in Wikipedia online encyclopedia as our dataset. Our proposed method can achieve highest accuracy with score about 98.8%. This experiment shows that our proposed method provides a robust system and good accuracy for classifying document into specified domain in preprocessing phase for domain specified ontology building.

Preliminary study of spam profile detection for social media using Markov Clustering: Case study on Javanese people

In this paper, we report our primary findings in detecting spam profiles on Facebook social media. As we already know, spam profiles on social media especially Facebook is often used as a place to spread spam. Spammers generally use wall post feature on Facebook to spread spam. In addition spammers are also targeting page on Facebook which is a community of many people. In this community, spammers can spread spam as desired. Based on that background, this paper aims is to provide solutions in order to reduce the impact from spammers by using Markov Clustering algorithm to detect spam profile. We used B-Cubed Metrics and F-Measure method to test how the clustering process performs. We obtained 220 profiles in Facebook which contains both normal profile and spam profile of Indonesian Facebook users at Javanese island. B-cubed metrics show accuracy 70% and 74% after applying majority voting to merge cluster into two cluster. F-Measure show accuracy 87% and 88% after applying majority voting. This indicates that algorithm has performed well.

Noun phrases extraction using shallow parsing with C4.5 decision tree algorithm for Indonesian Language ontology building

Ontology describes a set of concept or entity and each relation. Ontology as knowledge representation usually has a large structure because it can cover a wide area topics. Ontology building process is divided into two subprocesses, those are term extraction and relation formation. Term extraction in ontology building is done for extracting concept or entity before each relation is obtained. Main objective in this research is to extract noun phrases using shallow parsing algorithm based on C4.5 decision tree as candidate concept or term for ontology building process in Indonesian Text. One of the advantages of using shallow parsing is it can recover syntactic information efficiently and reliably from unrestricted text. For our dataset, we use Indonesian Language online newspapers for one month. Based on our experiments, it concludes that our proposed method can perform well for Indonesian Language noun phrase identification with average F-score 84.63%.

Noun ontology generation from Wikipedia article using Map Reduce with pattern based approach

Recently, data on the internet grows and it can be used as supporting information for human life. Wikipedia as an online encyclopedia provides many resources, data, and information on the internet. Main problem in our research is how to represent information from Indonesian Wikipedia article into some knowledge representation such as ontology. Ontology is a set of related concept and relation between those concepts. Ontology usually has a large and complex structure because ontology is made to cover a large area topic. Our approach in this ontology building is focused on hyponymy relation and meronymy relation. Our proposed method is using taxonomy template information in Wikipedia to extract hyponymy relation and some pattern to extract the meronymy relation. Our experiment shows that hyponymy relation can be extracted into 5038 relations. For our meronymy relation extraction process has 82.23% as the highest accuracy.

Locating nearest public facility using IDA∗ algorithm

Nowadays public transportation is needed given the current increasing transportation needs in the community because of the density of private vehicles that causes congestion everywhere. Given these reasons, currently people prefer to use public transportation to get to a location or perform daily life activities. But the problem is lack of information of public facilities and routes of available public transportation. This research will propose a system for searching public facility locations with implementation of IDA* algorithm. The input of this system is users location from a mobile device and the output will be a list of routes that can be viewed in a map.