Yeong Hyeon Gu

BLITE-SVR: New forecasting model for late blight on potato using support-vector regression

Various simple statistical methods have been used for the prediction of plant-disease epidemics. However, the need to develop a new model, reflecting many changed environmental factors and applicable to the Korean domestic farmhouse, has been raised. Given this point, we developed the potato late blight prediction model called BLITE-SVR, after which we predicted and verified the first date of occurrence with the data from 1976 to 1985 and from 2009 to 2012 through support-vector regression (SVR), a statistical method offering good performance. For the prediction model, we collected 13 kinds of weather data, including temperature, humidity, evaporation, and so on, which displayed very high correlation to the first date of the occurrence of late blight. The performance of BLITE-SVR has been evaluated through comparison with the conventional moving-average method that was previously used, as well as through pace regression and linear regression. The accuracy of prediction for the first date of occurrence was 64.3% by BLITE-SVR, thus showing a higher degree of accuracy compared with 42.9% by the conventional moving-average method, 42.9% by pace regression and 35.7% by linear regression. This study will enable farmers to match the targeted fungicide application to the time of greatest need and thereby achieve a reduction in chemical use.

Rules for Mining Comparative Online Opinions

The study of comparative online opinions is about sorting comparative sentences out of given sentences. This study, which is focused on the documents in Korean, may be the first of its kind in Korea although there have been a few of such studies in English spoken countries. In this study, 39 words –the most frequently used in the comparative sentences– were identified for the extraction of comparative sentences; especially, of the 39 words, this study is focused on the word ‘boda’, which is the most frequently occurring in Korean comparative sentences in identifying the rules for distinguishing the comparative sentences. The Korean word ‘boda’ is a proposition that has the same role as the English word, ‘than’; and if used as an adverb, ‘more’. In total, 11 rules were found in the observation of commodity review documents in blogs using the word ‘boda’. The study might be applied to a comparative search, on the internet, of a commodity or other object as well as an elementary technology of Opinion Mining.

Mining popular menu items of a restaurant from web reviews

We propose a novel method to mine popular menu items from online reviews. In order to extract popular menu items, a crawler that uses the wrapper on search web sites was used to collect online reviews, restaurant names, and menu items. Then, unnecessary posts were removed by using the patterns. Also, post frequency was used to find the most frequently appearing menu items from online reviews in order to select the most popular menu items. In the result, the total average accuracy was 0.900.

Performance analysis of combined descriptors for similar crop disease image retrieval

In this paper, combined image descriptors that can improve the performance of similar crop disease image retrieval system are suggested. When combining descriptors, the similarity between images is calculated using a single descriptor first. And, new similarity which corresponds to the combined descriptors is created by calculating the sum of image similarity corresponding to descriptors to be combined. Lastly, the image retrieval is carried out based on the distance value corresponding to the combined descriptors. The experiment was carried out with a total of 742 images of 3 crops including pear, grape and strawberry using the combined descriptors. As the experimental result, we discovered that using combined descriptors improved the system performance generally. And, we proved that a proper combination of descriptors varied for each crop and we found such combination. We also discovered that a combination of descriptors producing a high F-measure value of the system was different from a combination of descriptors having a higher probability that more accurate retrieval results would be outputted in the beginning of the screen. Therefore, proper combined descriptors should be selected according to actual system requirements.

User Preference Analysis and Visualization through the Browser History of Smart Devices

This paper studied the analysis of user preference and visualization through the browser history of smart devices. For the analysis of user preference, the browser history data which the user produced through smart devices in everyday life were automatically collected and transmitted to the server. Then using a web crawler the appropriate web pages were collected then classified by topics through a mechanical learning algorithm, and analyzed to determine which topic and category in the contents the user preferred. As a result of the experiment, the application of the Naïve Bayes algorithm yielded the best classifying capacity. And this experiment delivered the data through visualization to provide the user or the service provider reference. Through this test the experiment could classify the user character according to the consumption of contents by the smart device user, and understand the preference of the contents by the type of the user. Also, it could be utilized as an essential technique to provide personalized recommendation.

Recommender system architecture based on Mahout and a main memory database

In this study, we propose a news recommendation system architecture using a main memory database (DB) and Mahout. The user’s news preference rate is calculated automatically based on the time the user spends reading news items and their length. While existing systems also infer the user’s preferred fields, our system adjusts the volume and ratio of news stories using these categories. We collect web pages accessed by the user on a smart device and classify them using a naive Bayes classifier to determine the user’s preferred news categories. Collaborative filtering is then used to search for related news items read by others and to recommend news in a ratio consistent with the user’s preferred fields. Using a main memory DB, recommendations are computed 2.1 times faster than with a traditional DB when recommending from among 100,000 items; further, the more data used for recommendations, the bigger the speed difference between the proposed and traditional systems becomes.

Mining Comparative Sentences from Korean Text Documents Using Sentential Structure Analysis Combined with Machine Learning Techniques

This paper describes how to classify comparative and non-comparative sentences from Korean text documents by applying sentence structure analysis and machine learning analysis methods as well as analyze the performance of these methods. When applying the rules of sentence structure analysis, it is possible to classify comparative sentences with 100% precision. However, an error in the morpheme analyzing tool or orthography may lead to deteriorated accuracy of the classification for both comparative and non-comparative sentences. Rather than applying either the rules for sentence structure or machine learning methods, through this study, it has been proven, via experiments, that applying a combination of both can improve the accuracy of the analysis and this paper 1