• Journal of Korea Society of Industrial Information Systems
• /
• v.29 no.3
• /
• pp.1-12
• /
• 2024

In this paper, we fine-tuned KoAlpaca, a large-scale Korean language model, and implemented a blog text generation system utilizing it. Blogs on social media platforms are widely used as a marketing tool for businesses. We constructed training data of positive reviews through emotion analysis and refinement of collected sponsored blog texts and applied QLoRA for the lightweight training of KoAlpaca. QLoRA is a fine-tuning approach that significantly reduces the memory usage required for training, with experiments in an environment with a parameter size of 12.8B showing up to a 58.8% decrease in memory usage compared to LoRA. To evaluate the generative performance of the fine-tuned model, texts generated from 100 inputs not included in the training data produced on average more than twice the number of words compared to the pre-trained model, with texts of positive sentiment also appearing more than twice as often. In a survey conducted for qualitative evaluation of generative performance, responses indicated that the fine-tuned model's generated outputs were more relevant to the given topics on average 77.5% of the time. This demonstrates that the positive review generation language model for sponsored content in this paper can enhance the efficiency of time management for content creation and ensure consistent marketing effects. However, to reduce the generation of content that deviates from the category of positive reviews due to elements of the pre-trained model, we plan to proceed with fine-tuning using the augmentation of training data.

• Information Systems Review
• /
• v.22 no.1
• /
• pp.59-72
• /
• 2020

Artificial Intelligence is establishing itself as a familiar tool from an intractable concept. In this trend, financial sector is also looking to improve the problem of existing system which includes Fraud Detection System (FDS). It is being difficult to detect sophisticated cyber financial fraud using original rule-based FDS. This is because diversification of payment environment and increasing number of electronic financial transactions has been emerged. In order to overcome present FDS, this paper suggests 3 types of artificial intelligence models, Generative Adversarial Network (GAN), Deep Neural Network (DNN), and Convolutional Neural Network (CNN). GAN proves how data imbalance problem can be developed while DNN and CNN show how abnormal financial trading patterns can be precisely detected. In conclusion, among the experiments on this paper, WGAN has the highest improvement effects on data imbalance problem. DNN model reflects more effects on fraud classification comparatively.

• Food Science and Preservation
• /
• v.30 no.2
• /
• pp.262-271
• /
• 2023

The objective of this study is to compare and assess the effectiveness of real-time polymerase chain reaction (RT-PCR), loop-mediated isothermal amplification (LAMP), and the selective agar plate method for the detection of Salmonella spp. and Listeria monocytogenes in ready-to-eat (RTE) foods. In RTE foods, the detection performance of the three methods (RT-PCR [SureTect^TM kit and PowerChek^TM kit], LAMP [3M MDS], selective agar) were similar at 0-10, 10-50, 50-100, and 100- CFU/mL of Salmonella spp. and L. monocytogenes. We found that with RT-PCR, the Ct value of salad was significantly higher (p<0.05) than other RTE foods, indicating that fiber plays a critical role as an obstacle to the rapid detection of Salmonella spp. However, the Ct value displayed a mixed pattern according to the inoculation level of L. monocytogenes. The use of rapid detection kits and machines mostly depends on the user's choice, with accuracy, ease of use, and economy being the primary considerations. As an RT-PCR kit, SureTect^TM and PowerChek^TM showed high accuracy in detecting Salmonella spp. and L. monocytogenes in RTE foods, showing that they can replace the existing RT-PCR kits available. Additionally, LAMP also showed excellent detection performance, suggesting that it has the potential to be used as a food safety management tool.

• Korean Journal of Radiology
• /
• v.23 no.5
• /
• pp.555-565
• /
• 2022

Objective: To assess the efficacy and safety of ultrasound (US)-guided radiofrequency ablation (RFA) in patients with primary hyperparathyroidism (PHPT). Materials and Methods: This prospective study enrolled 39 participants (14 male, 25 female; mean age, 59.5 ± 15.3 [range, 18-87] years) between September 1, 2018, and January 31, 2021. All participants had parathyroid lesions causing PHPT, proven biochemically and through imaging. The imaging features of the PHPT nodules, including the shape, margin, size, composition, and location, were evaluated before treatment. Serum intact parathyroid hormone, calcium, and phosphorus levels; parathyroid nodule volume; and PHPT-related symptoms were recorded before and after treatment. We calculated the technical success, biochemical cure, and clinical cure rates for these patients. Complications were evaluated during and after the ablation. Results: Complete ablation was achieved in 38 of the 39 nodules in the 39 enrolled participants. All the patients were treated in one session. The technical success rate was 97.4% (38/39). The mean follow-up duration was 13.2 ± 4.6 (range, 6.0-24.9) months. At 6 and 12 months post-RFA, the biochemical cure rates were 82.1% (32/39) and 84.4% (27/32), respectively, and the clinical cure rates were 100% (39/39) and 96.9% (31/32), respectively. Only 2.6% (1/39) of the patients had recurrent PHPT. At 1, 3, 6, and 12 months after technically successful RFA, 44.7% (17/38), 34.3% (12/35), 15.8% (6/38), and 12.5% (4/32) of participants, respectively, had elevated eucalcemic parathyroid hormone levels. Recurrent laryngeal nerve paralysis occurred in 5.1% (2/39) of the patients, who recovered spontaneously within 1-3 months. Conclusion: US-guided RFA was effective and safe for PHPT patients. RFA may be an alternative treatment tool for patients who cannot tolerate or refuse to undergo surgery.

• Information Systems Review
• /
• v.21 no.4
• /
• pp.175-195
• /
• 2019

This study conducted both big data and netnography analysis to analyze consumer needs and behaviors of online consumer community. Big data analysis is easy to identify correlations, but causality is difficult to identify. To overcome this limitation, we used netnography analysis together. The netnography methodology is excellent for context grasping. However, there is a limit in that it is time and costly to analyze a large amount of data accumulated for a long time. Therefore, in this study, we searched for patterns of overall data through big data analysis and discovered outliers that require netnography analysis, and then performed netnography analysis only before and after outliers. As a result of analysis, the cause of the phenomenon shown through big data analysis could be explained through netnography analysis. In addition, it was able to identify the internal structural changes of the community, which are not easily revealed by big data analysis. Therefore, this study was able to effectively explain much of online consumer behavior that was difficult to understand as well as contextual semantics from the unstructured data missed by big data. The big data-netnography integrated model proposed in this study can be used as a good tool to discover new consumer needs in the online environment.

• Information Systems Review
• /
• v.19 no.2
• /
• pp.185-209
• /
• 2017

In modern society, crime records have been digitized. Digital information is difficult to distinguish from original information, but the former is easy to modulate. This situation explains the increasing importance of digital forensics. However, digital forensic has several inefficiencies because of the rapid development of technology, unclear jurisdiction, and tool errors. This study surveyed digital forensic specialists and derived the priority of domestic digital forensic issues by redefining 17 issues in digital forensics from Brungs-Jamieson study in Australia. The present study was divided into four groups, namely, police, government and public corporations, private companies, and legal groups. The study could compare and analyze comparative analysis of existing studies in Australia and the US. This study can also examine differences in the results of each group in Korea. Thus, the key issues in Korea were derived as "Requirements to 'Fire Up' Original." The differences of the three groups in terms of legal issues were then identified. This finding enables us to understand differences in priorities and importance between groups and countries.

• Information Systems Review
• /
• v.25 no.2
• /
• pp.205-227
• /
• 2023

Recently, Virtual Human(VH) has begun to appear in various media, not only on social media, but also in advertisements, music sources, and dramas. Virtual human has become a primary marketing tool for companies, but there also exist concerns when the companies do not disclose the identities of virtual humans. Accordingly, it is necessary to examine users' responses toward content that features virtual humans. This study aimed to examine how the exposure of virtual humans in the content affects users' perceptions. Therefore, the study defined the concept of 'VH Disclosure(VHD)', referring to the exposure of the virtual human's identity, and explored the impact of VH disclosure on attitude toward the ad (Hedonism, Utilitarianism, and Interestingness) and source credibility (Trustworthiness and Expertise). The study conducted an experimental survey with 302 respondents. Regardless of when the ad featured a VH or a human, the results showed that there was no significant difference between users' attitudes and source credibility. The results revealed that it was more effective to disclose the VH in social media feeds than directly reveal the VH's identity in the content. Therefore, this study utilizes a new concept of 'VH Disclosure(VHD)' to enhance the understanding of VH and contributes to establishing marketing strategies optimized for consumers in the creation of virtual human-related content.

• The Journal of Korean Medicine
• /
• v.44 no.4
• /
• pp.41-58
• /
• 2023

Objectives: This retrospective observational study aimed to investigate the efficacy and safety of Cheongpebaedok-tang, a traditional Korean herbal medicine, provided via telemedicine to patients with asymptomatic to mild COVID-19 in Korea. Methods: From February to April 2020, a retrospective analysis investigated COVID-19 patients treated via Korean telemedicine. The study involved asymptomatic to mild cases receiving Cheongpebaedok-tang more than three times, along with continuous Korean medicine care in convalescence. Diagnoses and treatment adhered to the telemedicine guidelines of the Association of Korean Medicine, with varied Cheongpebaedok-tang prescriptions based on symptom severity. Symptom evaluation involved a detailed assessment using a 15-item tool at initial and final sessions. Results: The study included 27 patients, with a mean age of 48.7 ± 2.3 years (mean ± standard error). Patients began self-administering oral Cheongpebaedok-tang for an average of 19.4 ± 1.8 days after the date of COVID-19 diagnosis confirmation and continued the medication for 15.8 ± 1.2 days. The reported side effects of the Cheongpebaedok-tang included palpitations (11.1%), insomnia (7.4%), dizziness (3.7%), and diarrhea (3.7%). All side effects disappeared after adjusting the prescription according to standard treatment guidelines. The occurrence of all COVID-19-related adverse symptoms, except fatigue and myalgia, decreased. Fatigue was the most common chronic symptom persisting after 6 months (51.9%), followed by ocular symptoms (37.0%) and sore throat (22.2%). Conclusions: This study implies Cheongpebaedok-tang may offer a potentially safe, symptom-alleviating approach for managing mild COVID-19 cases via telemedicine, although further comprehensive research is warranted.

• Journal of Music and Human Behavior
• /
• v.21 no.2
• /
• pp.31-49
• /
• 2024

The stress experienced by employees on international deployment has serious implications for their physical and mental health as well as their work performance. This study aimed to investigate the impact of resource-oriented music listening on work-related stress among employees on international deployment. The study recruited 10 participants of varying ages and occupations who had worked abroad for at least three years and conducted five personalized resource-oriented music listening sessions along with interviews. The Global Assessment of Recent Stress Scale (GARS) and Job Stress Scale (JSS) were used to measure stress levels before and after the intervention. Results showed that participants' work-related stress and perceived stress were significantly reduced following the music therapy intervention. The study also found that "awareness of own's life situation" and "self-acceptance of here and now emotional challenges" were factors that reduced regulatory stress among employees on international deployment. This study highlights the potential of resource-oriented music listening as a useful tool for managing workplace stress and promoting the well-being of expatriate employees.

• Journal of Cerebrovascular and Endovascular Neurosurgery
• /
• v.25 no.1
• /
• pp.50-61
• /
• 2023

Objective: To describe the roadmapping technique and our three-year experience in the management of intracranial aneurysms in the hybrid operating room. Methods: We analyzed all patients who underwent surgical clipping for cerebral aneurysms with the roadmapping technique from January 2017 to September 2019. We report demographic, clinical, and morphological variables, as well as clinical and radiological outcomes. We further describe three illustrative cases of the technique. Results: A total of 13 patients were included, 9 of which (69.2%) presented with subarachnoid hemorrhage, with a total of 23 treated aneurysms. All patients were female, with a mean age of 47.7 years (range 31-63). All cases were anterior circulation aneurysms, the most frequent location being the ophthalmic segment of the internal carotid artery (ICA) in 11 cases (48%), followed by posterior communicating in 8 (36%), and ICA bifurcation in 2 (8%). Intraoperative clip repositioning was required in 9 aneurysms (36%) as a result of the roadmapping technique in the hybrid operating room. There were no residual aneurysms in our series, nor reported mortality. Conclusions: The roadmapping technique in the hybrid operating room offers a complementary tool for the adequate occlusion of complex intracranial aneurysms, as it provides a real time fluoroscopic-guided clipping technique, and clip repositioning is possible in a single surgical stage, whenever a residual portion of the aneurysm is identified. This technique also provides some advantages, such as immediate vasospasm identification and treatment with intra-arterial vasodilators, balloon proximal control for certain paraclinoid aneurysms, and simultaneous endovascular treatment in selected cases during a single stage.