Jeongin Kim

Extracting User Interests on Facebook

Recently, through the rapid development of smart devices, Facebook has been publicly recognized as a representative social network service. Facebook has profile information that is the basis to form network relationship between people as well as closed information sharing between users. Also, social plug-in “Like” is a feature that Facebook includes. Current researches based on Facebook have a problem of not considering this “Like” feature. This paper has proposed the method of extracting user’s interest by using Term Frequency of nouns and “Likes.” “Posts” and “Likes” were collected through Facebook Open API. Collected “Posts” were preprocessed and Term Frequency of nouns was calculated. After calculating weights of user interests by using Term Frequency of nouns and “Likes,” user interests were extracted by higher ranked user interests weighting.

Analysis on Smartphone Related Twitter Reviews by Using Opinion Mining Techniques

Due to the recent popularization of smartphones, it has become easy to get connected on Social Network Service (SNS) which caused the proliferation on the amount of tweets on Twitter. Many studies have been proposed to discover valuable meanings from Twitter text messages by using opinion mining. However, these researches have a side effect that it is only focused on positive and negative in the limited category. In this paper, we will attempt to examine which factors could affect the users interests or preferences by analyzing and comparing smartphone product reviews which were posted on Twitter, in multilateral categories, by using opinion mining.

A Method for Normalizing Non-standard Words in Online Social Network Services: A Case Study on Twitter

Due to the big developments of Smartphone devices and on-line social network services, people can share diverse information about what they have been experienced during a day with no constrain to time or location. This fact has changed entire previous online system. We simply insert a query to search engine or OSNSs by using Smartphone devices. Because of this effectiveness, text data in OSNSs is getting bigger including many noisy data especially non-standard words. People are likely to type a text in short format such as abbreviation, acronym, and more when they using Smartphone to send a message to their friends in order to save time and data usages. As a result of these reasons, non-standard words on the web is extremely increasing so it has to be normalize into standard words in order to enhance performance of Natural Language Processing. When we analyze plain text data to extract semantic meaning, this nosy data has been ignore even though it has valuable information. In order to overcome this problem, we address a method for normalizing non-standard words in OSNSs, particularly for Twitter text data. We analyzed more than fifty million tweets which was collected by Stanford University and normalized non-standard words into standard English words by using diverse coefficient method such as dice, jacard, ochiai, sorgenfrei, and more. We finally conclude this paper by comparing those coefficient methods with our proposed one.

A semantic weighting method for document classification based on Markov logic networks

This paper proposes a semantic weighting method to classify textural documents. Human lives in the world where web documents have a great potential and the amount of valuable information has been consistently growing over the year. There is a problem that finding relevant web documents corresponding to what users want is more difficult due to the huge amount of web size. For this reason, there have been many researchers overcome this problem. The most important thing is document classification. All documents are composed of numerous words. Many classification methods have been extracted keywords from documents and then analyzed keywords pattern or frequency. In this paper, we propose Category Term Weight (CTW) using keywords from documents in order to enhance performance in document classification. CTW combines keywords frequency with semantic information. The frequency and semantic information have a great potential to find similarities between documents. That is why we calculates CTW from collection of training documents. After this step, CTW from unknown document and CTW in previous Category Term Database will be applied by designed Markov Logic Networks Model. Our designed MLNs Model and existing Naive-bayse Model will be compared by applied CTW. The experimental results shows the improvement of precision compare with the existing model.

Monitoring social relationship among Twitter users by using NodeXL

Recently, various studies have been studying for measuring social relationship among online users by using social network service and expressing relationship through network. However, these previous researches did not consider the current time or generate relationships among the users, so they cannot clearly express the real time relationships among the users. In this study, to solve such problems, we collected the tweets of the Twitter and constructed a database by extracting user ID and hash tag based on MapReduce algorithm. By utilizing this, Twitter user relationship network map is designed and established, and this is compared and evaluated against Mentionmap network. As a result, the Twitter Network Map generated by the proposed method was more effective than the existing Mentionmap which used twitter OpenAPI.

Implementation of 1:N Communication Model Using Serial Communication in an RF-Based Environment

The most important thing in the Internet of Things (IoT) is data transmission and reception between devices. Accordingly, various types of communication modules have been introduced. Currently, Bluetooth module is the most widely used but it is not suitable for 1:N communication since it performs 1:1 communication through pairing process. To overcome this limitation, the present paper proposes a model of data transmission and reception where 1:N communication can be done based on RF communication module using Arduino serial communication.

Correcting Misspelled Words in Twitter Text

The SNS became popularized by computer, mobile devices, and tablets that are accessible to the Internet. Among SNS, Twitter posts the words of short texts and, it shares information. Twitter texts are the optimal data to extract new information, but as it may contain the information within the limited number of words, there are various limitations. To improve accuracy of extracting information within Twitter texts, the process of calibrating misspelled words shall be taken in advance. In conventional studies to correct the misspelled words of Twitter texts, the relationship between misspelled words and correcting words was resolved by concerning the dependency of co-occurrence words with misspelled words within sentences and morphophonemic similarity, but since the frequency of co-occurrence words of misspelled words is not concerned, it has not resolved to correct misspelled words completely. In this paper, to correct misspelled words in Twitter texts, the use of the character n-gram method concerning spelling information and the word n-gram method concerning frequency of co-occurrence words are to be proposed.

Research for the Pattern Analysis of Individual Interest Using SNS Data: Focusing on Facebook

The number of global SNS users is rapidly increasing because SNS is executed by the device called Smartphone and enables the users to overcome time and space barrier with low cost. This would be a proof that the SNS is deeply involved in the communication between many people and related data is continuously increasing. Like this, data volume for SNS continues to increase due to the increasing users and the experts are doing the research considering data reusability. In particular, even though there are many researches to provide the individually customized information using the comments which exist in the SNS, there is a limitation to get the personal information due to the number of characters (approximately 150 words) which is a characteristic of SNS. However, most data which are written by the users still exist because of the characteristic of SNS and continuously stored data are considered to be the data which could sufficiently identify the characteristics of the individual users. Therefore in this paper, interested areas of the individuals are to be identified using the comments of individual users and their friends, and the patterns are to be analyzed in each interested area and finally, the analyzed data will be used for future research.

Modeling for context-aware healthcare service using ontology

Recently, as smart devices are equipped with various kinds of sensors and wireless network interface, many studies in the field of U-healthcare service have been in progress to provide smart device based medical applications to patients or healthcare providers. This paper is a variety of health-related Ontology available on the Internet and context inference-based intelligent healthcare information services. The purpose of this paper is to build context ontology for the users healthcare service environment by making a model and to provide real-time intelligent healthcare service which has high satisfaction to the user by defining inference rules based on this.