• Journal of Intelligence and Information Systems
• /
• v.25 no.4
• /
• pp.141-154
• /
• 2019

Rapid growth of internet technology and social media is progressing. Data mining technology has evolved to enable unstructured document representations in a variety of applications. Sentiment analysis is an important technology that can distinguish poor or high-quality content through text data of products, and it has proliferated during text mining. Sentiment analysis mainly analyzes people's opinions in text data by assigning predefined data categories as positive and negative. This has been studied in various directions in terms of accuracy from simple rule-based to dictionary-based approaches using predefined labels. In fact, sentiment analysis is one of the most active researches in natural language processing and is widely studied in text mining. When real online reviews aren't available for others, it's not only easy to openly collect information, but it also affects your business. In marketing, real-world information from customers is gathered on websites, not surveys. Depending on whether the website's posts are positive or negative, the customer response is reflected in the sales and tries to identify the information. However, many reviews on a website are not always good, and difficult to identify. The earlier studies in this research area used the reviews data of the Amazon.com shopping mal, but the research data used in the recent studies uses the data for stock market trends, blogs, news articles, weather forecasts, IMDB, and facebook etc. However, the lack of accuracy is recognized because sentiment calculations are changed according to the subject, paragraph, sentiment lexicon direction, and sentence strength. This study aims to classify the polarity analysis of sentiment analysis into positive and negative categories and increase the prediction accuracy of the polarity analysis using the pretrained IMDB review data set. First, the text classification algorithm related to sentiment analysis adopts the popular machine learning algorithms such as NB (naive bayes), SVM (support vector machines), XGboost, RF (random forests), and Gradient Boost as comparative models. Second, deep learning has demonstrated discriminative features that can extract complex features of data. Representative algorithms are CNN (convolution neural networks), RNN (recurrent neural networks), LSTM (long-short term memory). CNN can be used similarly to BoW when processing a sentence in vector format, but does not consider sequential data attributes. RNN can handle well in order because it takes into account the time information of the data, but there is a long-term dependency on memory. To solve the problem of long-term dependence, LSTM is used. For the comparison, CNN and LSTM were chosen as simple deep learning models. In addition to classical machine learning algorithms, CNN, LSTM, and the integrated models were analyzed. Although there are many parameters for the algorithms, we examined the relationship between numerical value and precision to find the optimal combination. And, we tried to figure out how the models work well for sentiment analysis and how these models work. This study proposes integrated CNN and LSTM algorithms to extract the positive and negative features of text analysis. The reasons for mixing these two algorithms are as follows. CNN can extract features for the classification automatically by applying convolution layer and massively parallel processing. LSTM is not capable of highly parallel processing. Like faucets, the LSTM has input, output, and forget gates that can be moved and controlled at a desired time. These gates have the advantage of placing memory blocks on hidden nodes. The memory block of the LSTM may not store all the data, but it can solve the CNN's long-term dependency problem. Furthermore, when LSTM is used in CNN's pooling layer, it has an end-to-end structure, so that spatial and temporal features can be designed simultaneously. In combination with CNN-LSTM, 90.33% accuracy was measured. This is slower than CNN, but faster than LSTM. The presented model was more accurate than other models. In addition, each word embedding layer can be improved when training the kernel step by step. CNN-LSTM can improve the weakness of each model, and there is an advantage of improving the learning by layer using the end-to-end structure of LSTM. Based on these reasons, this study tries to enhance the classification accuracy of movie reviews using the integrated CNN-LSTM model.

• Journal of Intelligence and Information Systems
• /
• v.26 no.1
• /
• pp.75-95
• /
• 2020

With the growth of social networks, various forms of SNS have emerged. Based on various motivations for use such as interactivity, information exchange, and entertainment, SNS users are also on the fast-growing trend. Facebook is the main SNS channel, and companies have started using Facebook pages as a public relations channel. To this end, in the early stages of operation, companies began to secure a number of fans, and as a result, the number of corporate Facebook fans has recently increased to as many as millions. from a corporate perspective, Facebook is attracting attention because it makes it easier for you to meet the customers you want. Facebook provides an efficient advertising platform based on the numerous data it has. Advertising targeting can be conducted using their demographic characteristics, behavior, or contact information. It is optimized for advertisements that can expose information to a desired target, so that results can be obtained more effectively. it rethink and communicate corporate brand image to customers through contents. The study was conducted through Facebook advertising data, and could be of great help to business people working in the online advertising industry. For this reason, the independent variables used in the research were selected based on the characteristics of the content that the actual business is concerned with. Recently, the company's Facebook page operation goal is to go beyond securing the number of fan pages, branding to promote its brand, and further aiming to communicate with major customers. the main figures for this assessment are Facebook's 'OK', 'Attachment', 'Share', and 'Number of Click' which are the dependent variables of this study. in order to measure the outcome of the target, the consumer's response is set as a key measurable key performance indicator (KPI), and a strategy is set and executed to achieve this. Here, KPI uses Facebook's ad numbers 'reach', 'exposure', 'like', 'share', 'comment', 'clicks', and 'CPC' depending on the situation. in order to achieve the corresponding figures, the consideration of content production must be prior, and in this study, the independent variables were organized by dividing into three considerations for content production into three. The effects of content material, content structure, and message styles on Facebook's user behavior were analyzed using regression analysis. Content materials are related to the content's difficulty, company relevance, and daily involvement. According to existing research, it was very important how the content would attract users' interest. Content could be divided into informative content and interesting content. Informational content is content related to the brand, and information exchange with users is important. Interesting content is defined as posts that are not related to brands related to interesting movies or anecdotes. Based on this, this study started with the assumption that the difficulty, company relevance, and daily involvement have an effect on the dependent variable. In addition, previous studies have found that content types affect Facebook user activity. I think it depends on the combination of photos and text used in the content. Based on this study, the actual photos were used and the hashtag and independent variables were also examined. Finally, we focused on the advertising message. In the previous studies, the effect of advertising messages on users was different depending on whether they were narrative or non-narrative, and furthermore, the influence on message intimacy was different. In this study, we conducted research on the behavior that Facebook users' behavior would be different depending on the language and formality. For dependent variables, 'OK' and 'Full Click Count' are set by every user's action on the content. In this study, we defined each independent variable in the existing study literature and analyzed the effect on the dependent variable, and found that 'good' factors such as 'self association', 'actual use', and 'hidden' are important. Could. Material difficulties', 'actual participation' and 'large scale * difficulties'. In addition, variables such as 'Self Connect', 'Actual Engagement' and 'Sexual Sexual Attention' have been shown to have a significant impact on 'Full Click'. It is expected that through research results, it is possible to contribute to the operation and production strategy of company Facebook operators and content creators by presenting a content strategy optimized for the purpose of the content. In this study, we defined each independent variable in the existing research literature and analyzed its effect on the dependent variable, and we could see that factors on 'good' were significant such as 'self-association', 'reality use', 'concernal material difficulty', 'real-life involvement' and 'massive*difficulty'. In addition, variables such as 'self-connection', 'real-life involvement' and 'formative*attention' were shown to have significant effects for 'full-click'. Through the research results, it is expected that by presenting an optimized content strategy for content purposes, it can contribute to the operation and production strategy of corporate Facebook operators and content producers.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.12 no.9
• /
• pp.3856-3868
• /
• 2011

Until recently, research trend in real estate has been focused on real estate market and the market analysis. But the studies on real estate training program development for real estate agents to improve their job performance are relatively short in numbers. Thus, this study shows empirical analysis of the needs for the training programs for real estate agents in Cheonan to improve their job performance. The results are as follows. First, in the survey of asking what educational contents they need in order to improve real estate agents' job performance, most of the respondents show their needs for the analysis of house's value, legal knowledge, real estate management, accounting, real estate marketing, and understanding of the real estate policy. This is because they are well aware that the best way of responding to the changing clients' needs comes from training programs. Secondly, asked about real estate marketing strategies, most of respondents showed their awareness of new strategies to meet the needs of clients. This is because new forms of marketing strategies including internet ads are needed in the field as the paradigm including Information Technology changes. Thirdly, asked about the need for real estate-related training programs, 92% of the respondents answered they need real estate education programs run by the continuing education centers of the universities. In addition, the survey showed their needs for retraining programs that utilize the resources in the local universities. Other than this, to have effective and efficient training programs, they demanded running a training system by utilizing the human resources of the universities under the name of the department of 'Real Estate Contract' for real estate agents' job performance. Fourthly, the survey revealed real estate management(44.2%) and real estate marketing(42.3%) is the most chosen contents they want to take in the regular course for improving real estate agents' job performance. This shows their will to understand clients' needs through the mind of real estate management and real estate marketing. The survey showed they prefer the training programs as an irregular course to those in the regular one. Despite the above results, this study chose subjects only in Cheanan and thus it needs to research more diverse areas. The needs of programs to improve real estate agents job performance should be analyzed empirically targeting the real estate agents not just in Cheonan but also cities like Pyeongchon, Ilsan and Bundang in which real estate business is booming, as well as undergraduate and graduate students whose major is real estate studies. These studies will be able to provide information to help develop the customized training programs by evaluating elements that real estate agents need in order to meet clients satisfaction and improve their job performance. Many variables of the program development learned through these studies can be incorporated in the curriculum of the real estate studies and used very practically as information for the development of the real estate studies in this fast changing era.

• Journal of Intelligence and Information Systems
• /
• v.19 no.1
• /
• pp.19-33
• /
• 2013

In this research, a proposed Dynamic Virtual Ontology using Tags (DyVOT) supports dynamic search of resources depending on user's requirements using tags from social web driven resources. It is general that the tags are defined by annotations of a series of described words by social users who usually tags social information resources such as web-page, images, u-tube, videos, etc. Therefore, tags are characterized and mirrored by information resources. Therefore, it is possible for tags as meta-data to match into some resources. Consequently, we can extract semantic relationships between tags owing to the dependency of relationships between tags as representatives of resources. However, to do this, there is limitation because there are allophonic synonym and homonym among tags that are usually marked by a series of words. Thus, research related to folksonomies using tags have been applied to classification of words by semantic-based allophonic synonym. In addition, some research are focusing on clustering and/or classification of resources by semantic-based relationships among tags. In spite of, there also is limitation of these research because these are focusing on semantic-based hyper/hypo relationships or clustering among tags without consideration of conceptual associative relationships between classified or clustered groups. It makes difficulty to effective searching resources depending on user requirements. In this research, the proposed DyVOT uses tags and constructs ontologyfor effective search. We assumed that tags are extracted from user requirements, which are used to construct multi sub-ontology as combinations of tags that are composed of a part of the tags or all. In addition, the proposed DyVOT constructs ontology which is based on hierarchical and associative relationships among tags for effective search of a solution. The ontology is composed of static- and dynamic-ontology. The static-ontology defines semantic-based hierarchical hyper/hypo relationships among tags as in (http://semanticcloud.sandra-siegel.de/) with a tree structure. From the static-ontology, the DyVOT extracts multi sub-ontology using multi sub-tag which are constructed by parts of tags. Finally, sub-ontology are constructed by hierarchy paths which contain the sub-tag. To create dynamic-ontology by the proposed DyVOT, it is necessary to define associative relationships among multi sub-ontology that are extracted from hierarchical relationships of static-ontology. The associative relationship is defined by shared resources between tags which are linked by multi sub-ontology. The association is measured by the degree of shared resources that are allocated into the tags of sub-ontology. If the value of association is larger than threshold value, then associative relationship among tags is newly created. The associative relationships are used to merge and construct new hierarchy the multi sub-ontology. To construct dynamic-ontology, it is essential to defined new class which is linked by two more sub-ontology, which is generated by merged tags which are highly associative by proving using shared resources. Thereby, the class is applied to generate new hierarchy with extracted multi sub-ontology to create a dynamic-ontology. The new class is settle down on the ontology. So, the newly created class needs to be belong to the dynamic-ontology. So, the class used to new hyper/hypo hierarchy relationship between the class and tags which are linked to multi sub-ontology. At last, DyVOT is developed by newly defined associative relationships which are extracted from hierarchical relationships among tags. Resources are matched into the DyVOT which narrows down search boundary and shrinks the search paths. Finally, we can create the DyVOT using the newly defined associative relationships. While static data catalog (Dean and Ghemawat, 2004; 2008) statically searches resources depending on user requirements, the proposed DyVOT dynamically searches resources using multi sub-ontology by parallel processing. In this light, the DyVOT supports improvement of correctness and agility of search and decreasing of search effort by reduction of search path.

• Journal of Intelligence and Information Systems
• /
• v.25 no.3
• /
• pp.117-140
• /
• 2019

As the aging phenomenon accelerates and various social problems related to the elderly of the vulnerable are raised, the need for effective elderly care solutions to protect the health and safety of the elderly generation is growing. Recently, more and more people are using Smart Toys equipped with ICT technology for care for elderly. In particular, log data collected through smart toys is highly valuable to be used as a quantitative and objective indicator in areas such as policy-making and service planning. However, research related to smart toys is limited, such as the development of smart toys and the validation of smart toy effectiveness. In other words, there is a dearth of research to derive insights based on log data collected through smart toys and to use them for decision making. This study will analyze log data collected from smart toy and derive effective insights to improve the quality of life for elderly users. Specifically, the user profiling-based analysis and elicitation of a change in quality of life mechanism based on behavior were performed. First, in the user profiling analysis, two important dimensions of classifying the type of elderly group from five factors of elderly user's living management were derived: 'Routine Activities' and 'Work-out Activities'. Based on the dimensions derived, a hierarchical cluster analysis and K-Means clustering were performed to classify the entire elderly user into three groups. Through a profiling analysis, the demographic characteristics of each group of elderlies and the behavior of using smart toy were identified. Second, stepwise regression was performed in eliciting the mechanism of change in quality of life. The effects of interaction, content usage, and indoor activity have been identified on the improvement of depression and lifestyle for the elderly. In addition, it identified the role of user performance evaluation and satisfaction with smart toy as a parameter that mediated the relationship between usage behavior and quality of life change. Specific mechanisms are as follows. First, the interaction between smart toy and elderly was found to have an effect of improving the depression by mediating attitudes to smart toy. The 'Satisfaction toward Smart Toy,' a variable that affects the improvement of the elderly's depression, changes how users evaluate smart toy performance. At this time, it has been identified that it is the interaction with smart toy that has a positive effect on smart toy These results can be interpreted as an elderly with a desire to meet emotional stability interact actively with smart toy, and a positive assessment of smart toy, greatly appreciating the effectiveness of smart toy. Second, the content usage has been confirmed to have a direct effect on improving lifestyle without going through other variables. Elderly who use a lot of the content provided by smart toy have improved their lifestyle. However, this effect has occurred regardless of the attitude the user has toward smart toy. Third, log data show that a high degree of indoor activity improves both the lifestyle and depression of the elderly. The more indoor activity, the better the lifestyle of the elderly, and these effects occur regardless of the user's attitude toward smart toy. In addition, elderly with a high degree of indoor activity are satisfied with smart toys, which cause improvement in the elderly's depression. However, it can be interpreted that elderly who prefer outdoor activities than indoor activities, or those who are less active due to health problems, are hard to satisfied with smart toys, and are not able to get the effects of improving depression. In summary, based on the activities of the elderly, three groups of elderly were identified and the important characteristics of each type were identified. In addition, this study sought to identify the mechanism by which the behavior of the elderly on smart toy affects the lives of the actual elderly, and to derive user needs and insights.

• The Korean Journal of Archival Studies
• /
• no.16
• /
• pp.303-339
• /
• 2007

"The damage incurred from forced mobilization under the Japanese Imperialism" means the life, physical, and property damage suffered by those who were forced to lead a life as soldiers, civilians attached to the military, laborers, and comfort women forcibly mobilized by the Japanese Imperialists during the period between the Manchurian Incident and the Pacific War. Up to the present time, every effort to restore the history on such a compulsory mobilization-borne damage has been made by the damaged parties, bereaved families, civil organizations, and academic circles concerned; as a result, on March 5, 2004, Disclosure act of Forced Mobilization under the Japanese Imperialism[part of it was partially revised on May 17, 2007]was officially established and proclaimed. On the basis of this law, the Truth Commission on Forced Mobilization under the Japanese Imperialism Republic of Korea[Compulsory Mobilization Commission hence after] was launched under the jurisdiction of the Prime Minister on November 10, 2004. Since February 1, 2005, this organ has begun its work with the aim of looking into the real aspects of damage incurred from compulsory mobilization under the Japanese Imperialism, by which making the historical truth open to the world. The major business of this organ is to receive the damage report and investigation of the reported damage[examination of the alleged victims and bereaved families, and decision-making], receipt of the application for the fact-finding & fact finding; fact finding and matters impossible to make judgment; correction of a family register subsequent to the damage judgement; collection & analysis of data concerning compulsory mobilization at home and from abroad and writing up of a report; exhumation of the remains, remains saving, their repatriation, and building project for historical records hall and museum & memorial place, etc. The Truth Commission on Compulsory Mobilization has dug out and collected a variety of records to meet the examination of the damage and fact finding business. As is often the case with other history of damage, the records which had already been made open to the public or have been newly dug out usually have their limits to ascertaining of the diverse historical context involved in compulsory mobilization in their quantity or quality. Of course, there may happen a case where the interested parties' story can fill the vacancy of records or has its foundational value more than its related record itself. The Truth Commission on Compulsory mobilization generated a variety of oral history records through oral interviews with the alleged damage-suffered survivors and puts those data to use for examination business, attempting to make use of those data for public use while managing those on a systematic method. The Truth Commission on compulsory mobilization-possessed oral history archives were generated based on a drastic planning from the beginning of their generation, and induced digital medium-based production of those data while bearing the conveniences of their management and usage in mind from the stage of production. In addition, in order to surpass the limits of the oral history archives produced in the process of the investigating process, this organ conducted several special training sessions for the interviewees and let the interviewees leave their real context in time of their oral testimony in an interview journal. The Truth Commission on compulsory mobilization isn't equipped with an extra records management system for the management of the collected archives. The digital archives are generated through the management system of the real aspects of damage and electronic approval system, and they plays a role in registering and searching the produced, collected, and contributed records. The oral history archives are registered at the digital archive and preserved together with real records. The collected oral history archives are technically classified at the same time of their registration and given a proper number for registration, classification, and keeping. The Truth Commission on compulsory mobilization has continued its publication of oral history archives collection for the positive use of them and is also planning on producing an image-based matters. The oral history archives collected by this organ are produced, managed and used in as positive a way as possible surpassing the limits produced in the process of investigation business and budgetary deficits as well as the absence of records management system, etc. as the form of time-limit structure. The accumulated oral history archives, if a historical records hall and museum should be built as regulated in Disclosure act of forced mobilization, would be more systematically managed and used for the public users.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.

• Korean Journal of Poultry Science
• /
• v.45 no.3
• /
• pp.219-228
• /
• 2018

This study was conducted twice to investigate egg purchase behavior and perception on animal welfare of Korean consumers. This study included women, who were the main decision makers and caretakers in the household, and men with one-person household. This survey was conducted with by the Computer Assisted Web Interview and Gang Survey methods. On the key considerations factor, the highest response rate was considered to be 'price', and the response rate of considering 'packing date' increased in the second survey. At a reasonable price based on 10 eggs, the response rate was the highest at 53.8% and 42.9% in both the first and second surveys and the appropriate price averages were 2,482 won and 2,132 won, respectively. The highest rate of purchase of egg consumers from 'Large Mart' followed by 'Medium sized supermarket' and 'Chain supermarket'. As for the awareness about animal welfare, the recognition ratio (73.5%) was higher in the result of the second survey than the first. The cognitive period of animal welfare was 59.0% before the insecticide egg crisis and 41.0% thereafter. Regarding whether or not they have ever seen an animal welfare certification mark and an animal welfare animal farm certification mark, 59.6% of respondents said that they saw it for the first time and 37.6% answered that they knew the animal welfare certification mark. On the animal welfare system, the 'free-range' response rate was the highest at 85.8%. The 'free-range' fit response decreased by 34.2%p, while the 'barn' and 'European type' fit response increased by 13.2%p and 24.1%p, respectively. The number of 'I have never seen' and 'I have ever eaten' responses to the recognition and eating experience of animal welfare certified eggs decreased while the number of those who answered 'Have ever seen' and 'Have eaten' increased. The answer of purchasing animal welfare certified eggs at department stores, organic farming cooperatives, and internet shopping malls was higher than that of buying conventional eggs. Of the total respondents, 92.0% were willing to purchase an animal welfare egg before the price was offered, but after offering the prices of animal welfare eggs, the intention to purchase was 62.7%, which was about 30%p lower than before. The reason for purchasing an animal welfare certified egg was the highest score of 71.0% for 'I think it is likely to be high in food safety', and 38.1% for 'I think the price is high' for lack of intention to purchase. In the sensory evaluation of animal welfare eggs, egg color and skin texture of conventional eggs were significantly higher than those of certified welfare eggs (P<0.05), and boiled eggs showed that egg whites of animal welfare certified eggs were more (P<0.05). As a result, the results of this study will contribute to the activation of the animal welfare certification system for laying hens by providing basic data on consumer awareness to animal welfare certified farmers.

• Journal of Intelligence and Information Systems
• /
• v.23 no.1
• /
• pp.95-108
• /
• 2017

Recently, AlphaGo which is Bakuk (Go) artificial intelligence program by Google DeepMind, had a huge victory against Lee Sedol. Many people thought that machines would not be able to win a man in Go games because the number of paths to make a one move is more than the number of atoms in the universe unlike chess, but the result was the opposite to what people predicted. After the match, artificial intelligence technology was focused as a core technology of the fourth industrial revolution and attracted attentions from various application domains. Especially, deep learning technique have been attracted as a core artificial intelligence technology used in the AlphaGo algorithm. The deep learning technique is already being applied to many problems. Especially, it shows good performance in image recognition field. In addition, it shows good performance in high dimensional data area such as voice, image and natural language, which was difficult to get good performance using existing machine learning techniques. However, in contrast, it is difficult to find deep leaning researches on traditional business data and structured data analysis. In this study, we tried to find out whether the deep learning techniques have been studied so far can be used not only for the recognition of high dimensional data but also for the binary classification problem of traditional business data analysis such as customer churn analysis, marketing response prediction, and default prediction. And we compare the performance of the deep learning techniques with that of traditional artificial neural network models. The experimental data in the paper is the telemarketing response data of a bank in Portugal. It has input variables such as age, occupation, loan status, and the number of previous telemarketing and has a binary target variable that records whether the customer intends to open an account or not. In this study, to evaluate the possibility of utilization of deep learning algorithms and techniques in binary classification problem, we compared the performance of various models using CNN, LSTM algorithm and dropout, which are widely used algorithms and techniques in deep learning, with that of MLP models which is a traditional artificial neural network model. However, since all the network design alternatives can not be tested due to the nature of the artificial neural network, the experiment was conducted based on restricted settings on the number of hidden layers, the number of neurons in the hidden layer, the number of output data (filters), and the application conditions of the dropout technique. The F1 Score was used to evaluate the performance of models to show how well the models work to classify the interesting class instead of the overall accuracy. The detail methods for applying each deep learning technique in the experiment is as follows. The CNN algorithm is a method that reads adjacent values from a specific value and recognizes the features, but it does not matter how close the distance of each business data field is because each field is usually independent. In this experiment, we set the filter size of the CNN algorithm as the number of fields to learn the whole characteristics of the data at once, and added a hidden layer to make decision based on the additional features. For the model having two LSTM layers, the input direction of the second layer is put in reversed position with first layer in order to reduce the influence from the position of each field. In the case of the dropout technique, we set the neurons to disappear with a probability of 0.5 for each hidden layer. The experimental results show that the predicted model with the highest F1 score was the CNN model using the dropout technique, and the next best model was the MLP model with two hidden layers using the dropout technique. In this study, we were able to get some findings as the experiment had proceeded. First, models using dropout techniques have a slightly more conservative prediction than those without dropout techniques, and it generally shows better performance in classification. Second, CNN models show better classification performance than MLP models. This is interesting because it has shown good performance in binary classification problems which it rarely have been applied to, as well as in the fields where it's effectiveness has been proven. Third, the LSTM algorithm seems to be unsuitable for binary classification problems because the training time is too long compared to the performance improvement. From these results, we can confirm that some of the deep learning algorithms can be applied to solve business binary classification problems.

• Journal of Intelligence and Information Systems
• /
• v.21 no.2
• /
• pp.191-204
• /
• 2015

According to the '2013 Performance Assessment Report on the Financial Program' from the National Assembly Budget Office, the unfilled recruitment ratio of Software(SW) Developers in South Korea was 25% in the 2012 fiscal year. Moreover, the unfilled recruitment ratio of highly-qualified SW developers reaches almost 80%. This phenomenon is intensified in small and medium enterprises consisting of less than 300 employees. Young job-seekers in South Korea are increasingly avoiding becoming a SW developer and even the current SW developers want to change careers, which hinders the national development of IT industries. The Korean government has recently realized the problem and implemented policies to foster young SW developers. Due to this effort, it has become easier to find young SW developers at the beginning-level. However, it is still hard to recruit highly-qualified SW developers for many IT companies. This is because in order to become a SW developing expert, having a long term experiences are important. Thus, improving job continuity intentions of current SW developers is more important than fostering new SW developers. Therefore, this study surveyed the job continuity intentions of SW developers and analyzed the factors associated with them. As a method, we carried out a survey from September 2014 to October 2014, which was targeted on 130 SW developers who were working in IT industries in South Korea. We gathered the demographic information and characteristics of the respondents, work environments of a SW industry, and social positions for SW developers. Afterward, a regression analysis and a decision tree method were performed to analyze the data. These two methods are widely used data mining techniques, which have explanation ability and are mutually complementary. We first performed a linear regression method to find the important factors assaociated with a job continuity intension of SW developers. The result showed that an 'expected age' to work as a SW developer were the most significant factor associated with the job continuity intention. We supposed that the major cause of this phenomenon is the structural problem of IT industries in South Korea, which requires SW developers to change the work field from developing area to management as they are promoted. Also, a 'motivation' to become a SW developer and a 'personality (introverted tendency)' of a SW developer are highly importantly factors associated with the job continuity intention. Next, the decision tree method was performed to extract the characteristics of highly motivated developers and the low motivated ones. We used well-known C4.5 algorithm for decision tree analysis. The results showed that 'motivation', 'personality', and 'expected age' were also important factors influencing the job continuity intentions, which was similar to the results of the regression analysis. In addition to that, the 'ability to learn' new technology was a crucial factor for the decision rules of job continuity. In other words, a person with high ability to learn new technology tends to work as a SW developer for a longer period of time. The decision rule also showed that a 'social position' of SW developers and a 'prospect' of SW industry were minor factors influencing job continuity intensions. On the other hand, 'type of an employment (regular position/ non-regular position)' and 'type of company (ordering company/ service providing company)' did not affect the job continuity intension in both methods. In this research, we demonstrated the job continuity intentions of SW developers, who were actually working at IT companies in South Korea, and we analyzed the factors associated with them. These results can be used for human resource management in many IT companies when recruiting or fostering highly-qualified SW experts. It can also help to build SW developer fostering policy and to solve the problem of unfilled recruitment of SW Developers in South Korea.