Natural Language Understanding Engine (LEA)


    Natural Language Understanding Engine LEA

    LEA (Language Engineering & Analysis) is a language analysis engine based on machine learning/deep learning that handles text analysis functions such as morpheme analysis, object name recognition, sentence structure analysis, and emotion analysis for non-structured data processing. In addition, natural language processing results are used to provide a higher level of analysis, such as understanding hidden intentions in sentences or identifying question types, which allows the system to understand the intentions for conversation processing and to understand the meaning of questions for analysis and deep Q/A. The natural language understanding engine (LEA) is the basic engine needed for other engines included in AI Suite to operate.

    The high-precision language analyzers that make up the natural language engine are applied by machine learning and deep learning (artificial neural network) technologies, and can be used to optimize the quality of each domain through large-scale language resources (large-capacity learning data, dictionaries and rules). Morpheme analyzer provides analysis quality of more than 98%, and syntax analysis and object name extractors provide the world best performance through parallel/distributed processing. LEA engine enables multilingual response in such languages as Korean, English, Japanese, and others, and is a natural language processing engine which could realize semantic analysis, Q&A, and dialogue systems by connecting with knowledge graph.


    < Natural Language Understanding Engine – LEA block diagram >

    Main Features

    High-quality natural language processing based on machine learning and deep learning
    Machine learning and deep learning (AI network) technologies are applied in high-precision language analyzers that make up LEA. Morpheme analysis and object name recognition based on the latest machine learning model Structural-SVM, positive / negative emotional analysis based on Latent Structural-SVM, and dependent sentence structure analysis in Transition-Based (Arc-Eager) Dependency Parsing together provide faster and higher performance than existing algorithms. In addition, Word Embedding makes it possible to apply deep learning to natural language processing.
    Ease of domain application
    Unlike other natural language engines where language processing involves common words (terms), the LEA engine can optimize the quality of each domain by using large language resources. It supports features of separately building large-capacity learning data to allow learning. In addition to the common dictionaries, dictionaries and rules specific to each domain are utilized. Thus, analysis can be tailored to the linguistic characteristics unique to such areas as healthcare, finance, and law.
    Meaning identification through the connection with knowledge graph
    The natural language understanding engine identifies not only natural language processing, such as morpheme analysis and sentence structure analysis, but also the objects and meanings of the analyzed words (features). Such meaning identification is made possible through connection with the knowledge base. It uses the knowledge graph to determine which objects do the words refer to in the actual knowledge, and tags (attaches a tag) them in a form that machine understands. The semantic information of knowledge graph is used as core information needed for artificial intelligence knowledge processing, such as in grasping the intention of spoken sentences in the dialogue processing or identifying the type and meaning of the question in Q&A processing.

    Main functions and specifications

    Natural language processing features
    The natural language understanding engine provides basic natural language processing features for non-structured text entered through morpheme analyzers, object name recognizers, and sentence structure analyzers.
    Intention analysis feature
    This feature analyzes not only the lexical meaning expressed in a sentence, but also the meaning of the sentence and its intention, thus proposing meaning-based classification results. Unlike natural language processing result that is sensitive to spacing or typos by using simple dictionaries or rules, this feature recursively reconstructs and re-analyzes sentences until optimal analysis is deducted. Through this process, in addition to correcting errors in an input sentence, it could provide a strong analysis result regarding errors in sentences users input in dialogue processing or Q&A.
    Question understanding features
    Natural language understanding engine provides not only simple natural language processing results but also semantic analysis results to understand user's questions in conversation processing or Q&A processing.
    The semantic object is identified by linking the natural language processing results of the input sentence with knowledge graph information. This feature also determines whether it is a declarative sentence or a question, and if it is a question, it categorizes what type of question it is. As such, by analyzing both semantic and syntactic knowledge information included in a sentence and deriving their results, the contents and intentions of the questions can be understood. The question understanding feature is a core function of the cognitive/understanding process for artificial intelligence services as it combines high-level language recognition technology with deep learning-based language / intention / knowledge learning technology with intention analysis.
    Dictionary management features
    It provides a dictionary management function in which major language dictionaries used by the engine are integrated and managed. The web-based integrated language dictionary management feature makes it easy to add and reflect language resources such as important terms used in a specific domain, or words and synonyms that need to be excluded during language processing. This provides analysis results which are customized based on users or domains, along with improved quality of language processing through periodic and continuous management.
    Untitled web

    Main engine screen

    Untitled web 1
    Untitled web 2