Ilmin Kim

An Experimental Evaluation of Short Opinion Document Classification Using A Word Pattern Frequency

An opinion mining technique which was developed from document classification in area of data mining now becomes a common interest in domestic as well as international industries. The core of opinion mining is to decide precisely whether an opinion document is a positive or negative one. Although many related approaches have been previously proposed, a classification accuracy was not satisfiable enough to applying them in practical applications. A opinion documents written in Korean are not easy to determine a polarity automatically because they often include various and ungrammatical words in expressing subjective opinions. Proposed in this paper is a new approach of classification of opinion documents, which considers only a frequency of word patterns and excludes the grammatical factors as much as possible. In proposed method, we express a document into a bag of words and then apply a learning algorithm using a frequency of word patterns, and finally decide the polarity of the document using a score function. Additionally, we also present the experiment results for evaluating the accuracy of the proposed method.

Intensity-based fiber optic pressure sensor with an Au/NiCr/Si3 N4/SiO2/Si3N4 diaphragm

Intensity based fiber optic pressure sensor has been fabricated using micromachining technology. The sensor consists of a multimode fiber bundle and a 100nm-Au/30nm- NiCr/150nm-Si3N4/300nm-SiO2/150nm- Si3N4 diaphragms supported by a micromachined frame-shape silicon substrate. The distance between the fiber bundle and the diaphragm of the pressure sensor is 900 micrometers . Three fiber optic pressure sensor having different size diaphragms were fabricated and used in experiments. The pressure sensitivities were 20.69 nW/kPa, 26.70 nW/kPa, and 39.33 nW/kPa for the pressure sensor with diaphragms of 3 X 3 mm2 area, 4 X 4 mm2 area, and 5 X 5 mm2 area.

Trajectory-Based Road-Geometry and Crash-Risk Estimation with Smartphone-Assisted Sensor Networks

As mobile devices came into wide use, it became practical to collect travel data in personal logs. Many studies have been conducted to extract meaningful information from this trend. In this study, we present a system for monitoring road-geometry and crash-risk estimation, based on trajectories created using a smartphone-aided sensor network. The proposed system consists of a number of node vehicles with smartphone applications for GPS data collection and a map server which aggregates the collected GPS trajectories and estimates road conditions. In order to estimate road geometry and crash risk information, the trajectories were segmented and categorized into groups according to their headings. Based on the processed trajectories, the geometry of the road section was estimated using the principal curve method. The crash risk of the road section was estimated from the constructed road geometry and the density map of the trajectories. Our system was evaluated using bicycle trajectories collected from segregated bicycle tracks in Seoul, Korea. Constructed geometry and crash-risk information of the track was compared with real track geometry and crash data. As a result, the estimated road geometry showed over 74% similarity and the calculated crash risk (61%) matched the real crash data.

Implementing a Web-Based Peer grading System and Experiment the Learning Effect

The popularity of online edcation technology has resulted in increase of interests on social learning network and online training aids for traditional class educations. In this paper, we designed and implemented a peer report grading system and studied the educational effect of the grading system. Instead of traditional one-sided evaluation done by professors, the system reduced the burden of grading and evaluating assinments and facilitated idea-sharing among students. We took advantage of web technology to maximize the effectiveness of peer feedback. In this paper, we concluded that the web based grading ststem showed better educational effect than traditional grading systems.

A Knowledge-Based Energy Management Model that Supports Smart Metering Networks for Korean Residential Energy Grids

Abstract The various energy management technologies that are required in order to deliver effective energy demand responses have resulted from the integrated use of digital technologies with energy grids. Therefore, the core technologies for smart metering infrastructure are regarded as a key issue in the design of future energy grids. The proposed knowledge-based model that supports advanced metering networks is capable of estimating energy consumption according to the characteristics of residential buildings. The energy consumption data is analyzed according to the residential building’s properties, which can significantly affect the energy consumption pattern. Therefore, appropriately designed models for energy consumption patterns with respect to identifying each energy consumption feature’s potential impact can be applied to create smart metering networks for future energy grid environments. This study introduces a knowledge-based model that considers both the energy and building management profiles. Then, case studies for the estimation of the energy consumption are presented. The proposed model could be effectively utilized in managing the energy demand response process with respect to market prices and residential energy shortages, and it provides a good reference for designing energy demand response strategies in Korean residential energy grid environments.