Although deep learning breakthroughs in NLP are based on learning distributed word representations by neural language models, these methods suffer from a classic drawback of unsupervised learning techniques. Furthermore, the performance of general-word embedding has been shown to be heavily task-dependent. To tackle this issue, recent researches have been proposed to learn the sentiment-enhanced word vectors for sentiment analysis. However, the common limitation of these approaches is that they require external sentiment lexicon sources and the construction and maintenance of these resources involve a set of complexing, time-consuming, and error-prone tasks. In this regard, this paper proposes a method of sentiment lexicon embedding that better represents sentiment word's semantic relationships than existing word embedding techniques without manually-annotated sentiment corpus. The major distinguishing factor of the proposed framework was that joint encoding morphemes and their POS tags, and training only important lexical morphemes in the embedding space. To verify the effectiveness of the proposed method, we conducted experiments comparing with two baseline models. As a result, the revised embedding approach mitigated the problem of conventional context-based word embedding method and, in turn, improved the performance of sentiment classification.

Attention-based long short-term memory network using sentiment lexicon embedding for aspect-level sentiment analysis in Korean

While clinical entity recognition mostly aims at electronic health records (EHRs), there are also the demands of dealing with the other type of text data. Automatic medical diagnosis is an example of new applications using a different data source. In this work, we are interested in extracting Korean clinical entities from a new medical dataset, which is completely different from EHRs. The dataset is collected from an online QA site for medical diagnosis. Bidirectional Encoder Representations from Transformers (BERT), which is one of the best language representation models, is used to extract the entities. A slightly modified version of BERT labeling strategy replaces the original labeling to enhance the separation of postpositions in Korean. A new clinical entity recognition dataset that we construct, as well as a standard NER dataset, have been used for the experiments. A pre-trained multilingual BERT model is used for the initialization of the entity recognition model. BERT significantly outperforms a character-level bidirectional LSTM-CRF, a benchmark model, in terms of all metrics. The micro-averaged precision, recall, and f1 of BERT are 0.83, 0.85 and 0.84, whereas that of bi-LSTM-CRF are 0.82, 0.79 and 0.81 respectively. The recall values of BERT are especially better than that of the other model. It can be interpreted that the trained BERT model could detect out of vocabulary (OOV) words better than bi-LSTM-CRF. The recently developed BERT and its WordPiece tokenization are effective for the Korean clinical entity recognition. The experiments using a new dataset constructed for the purpose and a standard NER dataset show the superiority of BERT compared to a state-of-the-art method. To the best of our knowledge, this work is one of the first studies dealing with clinical entity extraction from non-EHR data.

/pdf/korean-clinical-entity-recognition-from-diagnosis-text-using-2vk8h9n9fd.pdf

Korean clinical entity recognition from diagnosis text using BERT

Chinese word embeddings have recently attracted much attention in natural language processing (NLP). Existing researches learn Chinese word embeddings based on characters, radicals, components and stroke n-gram. Besides abovementioned features, Chinese characters also own structure and pinyin features. In this paper, we design feature substring, a super set of radicals, components and stroke n-gram with structure and pinyin information, to integrate stroke, structure and pinyin features of Chinese characters and capture the semantics of Chinese words. Based on the feature substring, we propose a novel method ssp2vec to predict the contextual words based on the feature substrings of the target words for learning Chinese word embeddings. It is based on our observation that exploiting the morphological information (stroke and structure) and the phonetic information (pinyin) is crucial for capturing the meanings of Chinese words. Meanwhile, the phonetic information (pinyin) can assist the model to distinguish Chinese words. Experimental results on word analogy, word similarity, text classification and named entity recognition tasks show that the proposed method obtains better results than state-of-the-art approaches.

Learning Chinese Word Embeddings from Stroke, Structure and Pinyin of Characters

The Korean Language: Structure, Use and Context

The recent surge of text-based online counseling applications enables us to collect and analyze interactions between counselors and clients. A dataset of those interactions can be used to learn to automatically classify the client utterances into categories that help counselors in diagnosing client status and predicting counseling outcome. With proper anonymization, we collect counselor-client dialogues, define meaningful categories of client utterances with professional counselors, and develop a novel neural network model for classifying the client utterances. The central idea of our model, ConvMFiT, is a pre-trained conversation model which consists of a general language model built from an out-of-domain corpus and two role-specific language models built from unlabeled in-domain dialogues. The classification result shows that ConvMFiT outperforms state-of-the-art comparison models. Further, the attention weights in the learned model confirm that the model finds expected linguistic patterns for each category.

Conversation Model Fine-Tuning for Classifying Client Utterances in Counseling Dialogues

Research on distributed word representations is focused on widely-used languages such as English. Although the same methods can be used for other languages, language-specific knowledge can enhance the accuracy and richness of word vector representations. In this paper, we look at improving distributed word representations for Korean using knowledge about the unique linguistic structure of Korean. Specifically, we decompose Korean words into the jamo-level, beyond the character-level, allowing a systematic use of subword information. To evaluate the vectors, we develop Korean test sets for word similarity and analogy and make them publicly available. The results show that our simple method outperforms word2vec and character-level Skip-Grams on semantic and syntactic similarity and analogy tasks and contributes positively toward downstream NLP tasks such as sentiment analysis.

/pdf/subword-level-word-vector-representations-for-korean-47d0e0m6kg.pdf

Yongseok Cho

Papers

Subword-level Word Vector Representations for Korean