Most countries provide veterans with various benefits to reward their sacrifice. Unfortunately, many veterans have failed to prove their status due to loss of military records. Thus, some governments allow the verification of those veterans through "buddy statements" obtained from the people who can vouch for the buddy's participation in the war. However, it is still challenging for veterans to find guarantors directly. With this background, we suggest to utilizing historical war records of combined operations to increase the pool of potential guarantors for the buddy statements. However, a combined operation network among troops can have missing edges and perturbations on attributes of the troop due to inaccurate information. In this study, we learn from some recorded interactions which might be incomplete and noisy, and predict missing linkages among the troops that might have interacted together in the war, by proposing Robust-SEAL (learning from Subgraphs, Embeddings, and Attributes for Link prediction). It combines two Graph Neural Network (GNN) architectures: robust Graph Convolutional Network which considers the uncertainty of node attributes with a probabilistic approach, and SEAL which improves the expressive power of the GNN with a labeling trick. Our proposed approach was applied to Korean War data with perturbations. For experimentations, we hid some actual interactions and found that Robust-SEAL restores missing interactions better than other GNN-based baselines.

Digital therapeutics (DTx) companies are making efforts to expand their consumer base in the growing market. To this end, many of them reposition their existing products to identify new target diseases. In this study, we provide a link prediction framework for DTx product repositioning on the basis of a multiplex disorder network by integrating multiple data sources, such as associated genes between disorders, shared drugs, and patents of treatment protocols. To capture the disorders' latent features, both random-walk-based and deep-learning-based graph embedding methods are applied to transform the graph structure into vectors. Consequently, new indications are suggested for DTx products based on the cosine similarity between original and candidate disorders. Our framework was applied to five psychiatric DTx products to determine new target disorders that have the highest treatment potential for each product. Therefore, the study results are expected to assist DTx firms in entering the novel target market with low risk within a short period. Moreover, the applicability of DTx products to a wider variety of disorders may increase access to overall patient groups and gradually improve public health.

Many algorithms for searching the main paths according to a subfield technology have been suggested using patent citation networks. However, the process by which the specific field is divided into subfields is problematic, in the sense that citation information among individual patents in different subfields is lost. If convergence technologies are common among different subfields, this problem can result in an inappropriate main path being found. To resolve this problem, we propose a new algorithm capable of extracting the main paths for each subfield of a technology by using the International Patent Classification (IPC) codes of patents. The proposed algorithm is applied to the waste management field to allow core technologies in the subfields of waste disposal, waste treatment, and the reuse of waste materials to be found. Finally, a sensitivity analysis is performed by considering the standard of the component ratio along a specific path to demonstrate the robustness of the proposed algorithm in the waste management field.

This study investigates recommendations for interdisciplinary research collaboration based on the weak ties theory. Contents-based features are proposed to recommend interdisciplinary collaboration considering that some researchers who have shown a preference for interdisciplinary collaboration could be connected even if they have dissimilar research profiles. Therefore, we inferred the preference of interdisciplinary research collaboration for every researcher, and considered features such as highlighting dissimilar researchers depending on their preferences. The features are designed to have typical similarity measures when the researchers do not prefer interdisciplinary research collaboration. We evaluated our proposed features with the baseline features of patent application datasets and the former methods outperformed the latter methods.

Various classification algorithms became available due to a surge of interdisciplinary research interests in the areas of data mining and knowledge discovery. We develop a statistical meta-model which compares the classification performances of several algorithms in terms of data characteristics. This empirical model is expected to aid decision making processes of finding the best classification tool in the sense of providing the minimum classification error among alternatives.