In recent times, with the development of science and technology, new technologies have been rapidly emerging, and innovators are making efforts to acquire intellectual property rights to preserve their competitive advantage as well as to enhance innovative competitiveness. As a result, the number of patents being acquired increases exponentially every year, and the social and economic ripple effects of developed technologies are also increasing. Now, innovators are focusing on evaluating existing technologies to develop more valuable ones. However, existing patent analysis studies mainly focus on discovering core technologies amongst the technologies derived from patents or analyzing trend changes for specific techniques; the analysis of innovators who develop such core technologies is insufficient. In this paper, we propose a model for analyzing the technical inventions of applicants based on patent classification systems such as international patent classification (IPC) and cooperative patent classification (CPC). Through the proposed model, the common invention patterns of applicants are extracted and used to analyze their technical inventions. The proposed model shows that patent classification systems can be used to extract the trends in applicants’ technological inventions and to track changes in their innovative patterns.

A number of sensor devices are widely distributed and used today owing to the accelerated development of IoT technology. In particular, this technological advancement has allowed users to carry IoT devices with more convenience and efficiency. Based on the IoT sensor data, studies are being actively carried out to recognize the current situation or to analyze and predict future events. However, research for existing smart healthcare services is focused on analyzing users’ behavior from single sensor data and is also focused on analyzing and diagnosing the current situation of the users. Therefore, a method for effectively managing and integrating a large amount of IoT sensor data has become necessary, and a framework considering data interoperability has become necessary. In addition, an analysis framework is needed not only to provide the analysis of the users’ environment and situation from the integrated data, but also to provide guide information to predict future events and to take appropriate action by users. In this paper, we propose a prescriptive analysis framework using a 5W1H method based on CKAN cloud. Through the CKAN cloud environment, IoT sensor data stored in individual CKANs can be integrated based on common concepts. As a result, it is possible to generate an integrated knowledge graph considering interoperability of data, and the underlying data is used as the base data for prescriptive analysis. In addition, the proposed prescriptive analysis framework can diagnose the situation of the users through analysis of user environment information and supports users’ decision making by recommending the possible behavior according to the coming situation of the users. We have verified the applicability of the 5W1H prescriptive analysis framework based on the use case of collecting and analyzing data obtained from various IoT sensors.

In traditional web crawling, all web pages crawled are first stored to databases. As a result, this approach can store unnecessary web pages and requires additional running time for the construction of a sentiment dictionary in a particular domain because sentiment words should be identified by scanning all web pages in the database. To address these problems, we first define the sentiment-aware web crawling problem and then propose two hash-based methods for the implementation. One is based on hash join and the other is bucket-sorted hash join. In particular, we propose a novel bucket-sorted hash join for the efficient sentiment-aware web crawling method. Our experimental results show that the proposed web crawling method using bucket-sorted hash join outperforms existing web crawling methods by significantly reducing the running time and storage space. In the proposed method, the time taken to execute the sentiment-aware task per web page is 0.016 seconds and the database space can be saved by 59% compared to the existing web crawling methods.

기존의 대학 평판도 평가 연구는 높은 비용과 감성의 편차가 발생하는 문제점을 지닌다. 따라서 이 논문에서는 이전 연구의 감성 사전 확장을 통한 대학 평판도 평가를 제안한다. 이를 위해, 우선 풍부한 데이터를 이용하여 기존의 감성 사전을 확장하고 군집화를 이용해 주제를 추출한다. 이러한 정보를 이용하여 대학의 평판도를 평가하고 추이를 분석한다. 연구 결과를 활용하여 기존의 설문 조사 기반 대학평가의 문제점을 개선할 수 있으며, 데이터의 크기와 대학 평판도의 관계를 파악할 수 있다. 또한, 주제별 대학평판의 개선을 위한 기초 자료로 활용될 수 있다.

사전 학습 언어 모델은 대용량 텍스트 데이터를 학습에 사용하여 다양한 자연어 처리 분야에서 우수한 성능을 보인다. 단어의 문맥 정보를 반영한 임베딩 기법들은 사전 학습 모델의 성능에 중요한 역할을 한다. 그렇지만 문맥에 출현한 명시적 단어들의 문맥 정보만 임베딩에 사용되고 있으므로 단어들이 가질 수 있는 다양한 관계 정보를 동시에 임베딩에 반영하기 위한 임베딩 기법이 필요하다. 본 논문에서는 지식 임베딩 기법을 통해 문장에 출현한 개체가 트리플에서 주어 개체인 경우뿐만 아니라 목적어 개체인 경우까지 포함시켜 출현 개체가 가질 수 있는 지식 정보를 확장하여 임베딩에 반영하는 확장 방법을 제안한다. 제안 모델의 성능 평가결과 기존 지식 기반 사전 학습 모델인 CoLAKE보다 우수한 성능을 보인다. 따라서 지식 임베딩을 통한 사전학습 모델의 성능 향상을 통해 응용 분야 문제 해결에 도움을 줄 것으로 기대한다.

Entity name recognition is a part of information extraction that extracts entity names from documents and classifies the types of extracted entity names. Entity name recognition technologies are widely used in natural language processing, such as information retrieval, machine translation, and query response systems. Various deep learning-based models exist to improve entity name recognition performance, but studies that compared and analyzed these models on Korean data are insufficient. In this paper, we compare and analyze the performance of CRF, LSTM-CRF, BiLSTM-CRF, and BERT, which are actively used to identify entity names using Korean data. Also, we compare and evaluate whether embedding models, which are variously used in recent natural language processing tasks, can affect the entity name recognition model