Jinhwa Kim

Combining models from neural networks and inductive learning algorithms

The knowledge-based artificial neural network (KBANN) is composed of phases involving the expression of domain knowledge, the abstraction of domain knowledge at neural networks, the training of neural networks, and finally, the extraction of rules from trained neural networks. The KBANN attempts to open up the neural network black box and generates symbolic rules with (approximately) the same predictive power as the neural network itself. An advantage of using KBANN is that the neural network considers the contribution of the inputs towards classification as a group, while rule-based algorithms like C5.0 measure the individual contribution of the inputs one at a time as the tree is grown. The knowledge consolidation model (KCM) combines the rules extracted using KBANN (NeuroRule), frequency matrix (which is similar to the Naive Bayesian technique), and C5.0 algorithm. The KCM can effectively integrate multiple rule sets into one centralized knowledge base. The cumulative rules from other single models can improve overall performance as it can reduce error-term and increase R-square. The key idea in the KCM is to combine a number of classifiers such that the resulting combined system achieves higher classification accuracy and efficiency than the original single classifiers. The aim of KCM is to design a composite system that outperforms any individual classifier by pooling together the decisions of all classifiers. Another advantage of KCM is that it does not need the memory space to store the dataset as only extracted knowledge is necessary in build this integrated model. It can also reduce the costs from storage allocation, memory, and time schedule. In order to verify the feasibility and effectiveness of KCM, personal credit rating dataset provided by a local bank in Seoul, Republic of Korea is used in this study. The results from the tests show that the performance of KCM is superior to that of the other single models such as multiple discriminant analysis, logistic regression, frequency matrix, neural networks, decision trees, and NeuroRule. Moreover, our model is superior to a previous algorithm for the extraction of rules from general neural networks.

Enhancement of User Understanding and Service Value Using Online Reviews

The Web has become an excellent source for gathering consumer opinions. There are now numerous Web sites containing such opinions, e.g., customer reviews of products, forums, discussion groups, and blogs. This paper focuses on online customer reviews of products. It makes some contributions. Especially it proposes minimalism and chunking framework for analyzing and comparing consumer opinions of competing products. Users are able to clearly see the strengths and weaknesses of each product in the minds of consumers in terms of various product features. This comparison is useful to both potential customers and product manufacturers. For a product manufacturer, the comparison enables it to easily gather marketing intelligence and product benchmarking information. In this paper, we only focus on mining opinion/product features that the reviewers have commented on. Five types of online review presentations are presented to mine such features. Our experimental results show that these techniques are useful to identify customers` opinions and trends.

A knowledge integration model for the prediction of corporate dividends

Dividend is one of essential factors determining the value of a firm. According to the valuation theory in finance, discounted cash flow (DCF) is the most popular and widely used method for the valuation of any asset. Since future dividends play a key role in the pricing of a current firm value by DCF, it is natural that the accurate prediction of future dividends should be most important work in the valuation. Although the dividend forecasting is of importance in the real world for the purpose of investment and financing decision, it is not easy for us to find good theoretical models which can predict future dividends accurately except Marsh and Merton [Marsh, T. A., & Merton, R. C. (1987). Dividend behavior for the aggregate stock market. Journal of Business, 60 (1), 1-40.] model. Thus, if we can develop a better method than Marsh and Merton (1987) in the prediction of future dividends, it can contribute significantly to the improvement of the pricing model of a firm value. Therefore, the most important goal of this study is to develop a better model by applying artificial intelligence techniques than Marsh and Merton (1987). The effectiveness of our approach was verified by the experiments comparing with Marsh and Merton model, Neural Networks, and CART approaches.

A Study on Characteristics of Eco-friendly Behaviors using Big Data: Focusing on the Customer Sales Data of Green Card

As part of a policy to address climate change and pollution problem, the government introduced a green credit card scheme in order to motivate pro-environmental behaviors in July 2011. It is important to present the specific ways to facilitate pro-environmental behaviors using the consumer behavior pattern data. This study was a result of data from total fifty seven thousands customer purchasing history data of green credit card to be created for the 3 months from January to March 2015. As the analysis process is put in to operation the analysis of the purchasing customer`s profile firstly, and the second come into association analysis to consider the buying associations for green products purchasing networks, the third estimate the useful parameters to affect the customer`s pro-environmental behavior and customer characteristics. It shows that royal customers are from 30 to 40 years old and their incomes are from 30 million won to 40 million won. Especially, they live in Daegu, Gyeonggi, and Seoul.

A Genetic Multi-Agent Rule Induction System for Stream Data

Many data mining algorithms are not capable of working effectively with very large stream data sets. Todays, organizations are building massive amounts of Internet-related stream data they collect, process, and store. Organizations want to mine effectively large stream data sets. But existing data mining algorithms have many critical problems. Storage management, increased run time, complexity of algorithms is the examples. This study constructs a new stream data mining algorithms, and builds knowledge base from very large stream data sets with genetic algorithm and rule induction system. Unlike exiting methods that build knowledge from stream data sets, genetic multi-agent rule induction system builds knowledge from the large stream data sets and then significantly improves prediction and classification accuracy.

A Knowledge Integration Model for Corporate Dividend Prediction

Dividend is one of essential factors determining the value of a firm. According to the valuation theory in finance, discounted cash flow (DCF) is the most popular and widely used method for the valuation of any asset. Since dividends play a key role in the pricing of a firm value by DCF, it is natural that the accurate prediction of future dividends should be most important work in the valuation. Although the dividend forecasting is of importance in the real world for the purpose of investment and financing decision, it is not easy for us to find good theoretical models which can predict future dividends accurately except Marsh and Merton (1987) model. Thus, if we can develop a better method than Marsh and Merton in the prediction of future dividends, it can contribute significantly to the enhancement of a firm value. Therefore, the most important goal of this study is to develop a better method than Marsh and Merton model by applying artificial intelligence techniques. The effectiveness of our approach was verified by the experiments comparing with Marsh and Merton model, Neural Networks, and CART approaches.