• Journal of Intelligence and Information Systems
• /
• v.28 no.2
• /
• pp.127-146
• /
• 2022

Recently, as word embedding has shown excellent performance in various tasks of deep learning-based natural language processing, researches on the advancement and application of word, sentence, and document embedding are being actively conducted. Among them, cross-language transfer, which enables semantic exchange between different languages, is growing simultaneously with the development of embedding models. Academia's interests in vector alignment are growing with the expectation that it can be applied to various embedding-based analysis. In particular, vector alignment is expected to be applied to mapping between specialized domains and generalized domains. In other words, it is expected that it will be possible to map the vocabulary of specialized fields such as R&D, medicine, and law into the space of the pre-trained language model learned with huge volume of general-purpose documents, or provide a clue for mapping vocabulary between mutually different specialized fields. However, since linear-based vector alignment which has been mainly studied in academia basically assumes statistical linearity, it tends to simplify the vector space. This essentially assumes that different types of vector spaces are geometrically similar, which yields a limitation that it causes inevitable distortion in the alignment process. To overcome this limitation, we propose a deep learning-based vector alignment methodology that effectively learns the nonlinearity of data. The proposed methodology consists of sequential learning of a skip-connected autoencoder and a regression model to align the specialized word embedding expressed in each space to the general embedding space. Finally, through the inference of the two trained models, the specialized vocabulary can be aligned in the general space. To verify the performance of the proposed methodology, an experiment was performed on a total of 77,578 documents in the field of 'health care' among national R&D tasks performed from 2011 to 2020. As a result, it was confirmed that the proposed methodology showed superior performance in terms of cosine similarity compared to the existing linear vector alignment.

• Archives of design research
• /
• v.20 no.1 s.69
• /
• pp.65-78
• /
• 2007

The mobile phone and wireless network, location based technology and other newly introduced technologies and communication media gave birth to the new terminology "ubiquitous" and are changing our daily life. Influence of such technologies and communication media is not an exception in the arts. New media art pieces using these technologies are increasing, and taking on the characteristics of public art within a wider scope of a city as a backdrop, beyond the traditional boundaries of art galleries. Of such art, locative media art using locative media has a closer relationship with city space than any other form of an, and makes various attempts to allow the spectator to reinterpret and experience city space and induce communication. These characteristics of locative media art can be considered as a method that can solve quality problems of the city space, especially the loss of the sense of place and the absence of communication. is one such locative media project with a purpose of solving quality problems of city space, especially the recovery of commercial sites and inducing communication. This project uses the paths of the city as its canvas, movement of people as its brush, the color of the roads as its pallet, and by allowing the partakers to draw paths of their own and to share their paths with others. People are encouraged to share stories about their paths. The project proceeds using barcodes that are frequently used commercially. When users wish to create their own place, they can enter their place and colors of their choice using input devices installed in the city space. Paths that are created through such a process will be displayed in public areas throughout the city, shared with others, and can create and share a stories about the city using on/off-line media.

• Korean Journal of Ecology and Environment
• /
• v.53 no.4
• /
• pp.380-389
• /
• 2020

Research on diagnosing and predicting the response of ecosystems caused by environmental changes such as artificial disturbance and climate change is emerging as the most important issue of biodiversity and ecosystem researches. This study aims to clean, standardize, and provide the results of National Ecosystem Survey which should be considered fundamentally in diagnosing and predicting ecosystem changes in the form of dataset. To refine and clean the dataset we developed a simple verification program based on the fifth National Ecosystem Survey Guideline and applied that program to the data from the second (1997~2005), third (2006~2013) and fourth (2014~2018) National Ecosystem Survey. Data quality control processes were implemented including (1) standardization of terminology, (2) similar data table integration, (3) unnecessary attribute and error elimination, (4) unification of different input items, (5) data arrangement in codes, and (6) code mapping for input items. These approaches and methods are the first attempt propose an option for ecological data standardization in Korea. The standardized dataset of National Ecosystem Survey in Korea will be easily accessible, reusable for both researchers and public. In addition, we expect it will contribute to the establishment of diverse environmental policies concerning environmental assessments, habitat conservation, prediction of endangered species distribution and ecological risks due to climate change. The dataset through this study is open freely online via EcoBank (nie-ecobank.kr) which is the first ecological information portal system in Korea developed by National Institute of Ecology.