Abstract: A method of retrieving a document according to an embodiment of the present application includes: acquiring a user retrieval query; calculating a user inquiry vector in a unit of sentence from the user retrieval query and acquiring a first document candidate group based on similarity between the calculated user inquiry vector and an embedding vector of a document stored in a retrieval database; acquiring a second document candidate group based on similarity between a text included in the user retrieval query and a text of the document stored in the retrieval database; and determining a summarization target document based on the first document candidate group and the second document candidate group.

Abstract: An unstructured document analysis method according to an embodiment includes: operations of acquiring unstructured document data including font characteristic data and document structure data, extracting text included in the unstructured document data on the basis of the font characteristic data or the document structure data, classifying the extracted text into a pre-classified item using a trained neural network model, acquiring a content query related to the content included in the unstructured document data and associated with the pre-classified item, and generating an answer to the content query on the basis of the extracted text classified into the item.

Abstract: Disclosed herein is a search method performed by an electronic device, including: receiving the plurality of similar question candidates selected by the server using a first similarity analysis model and question vectors corresponding to the plurality of similar question candidates from the server; selecting one or more similar questions from the plurality of similar question candidates by the electronic device using a second similarity analysis model; receiving the responses to the one or more similar questions from the server; and providing the one or more similar questions and the responses to the similar question together to the user.

Abstract: The present invention relates to a context-based QA generation architecture for generating diverse QA pairs from a single context. The context-based QA generation architecture includes a latent variable generating network, an answer generating network and a question generating network. The latent variable generating network comprises multiple Bi-LSTM encoders encode the a first context, the a first question and the a first answer to generate a first context vector, a first question vector and a first answer vector, respectively, a first Multi-Layer Perceptron (MLP) generate a first question latent variable based on the first context vector and the first question vector, and a second MLP generate a first answer latent variable based on the first question latent variable and the first answer vector. The answer generating network and the question generating network are trained based on the first context, the first question latent variable and the first answer latent variable.

Abstract: The invention relates to a method and a system for improving performance of text summarization and has an object of improving performance of a technique for generating a summary from a given paragraph. According to the invention to achieve the object, a method for improving performance of text summarization includes: calculating a first likelihood of each of a plurality of nodes included in a graph corresponding to a natural language-based context; calculating a second likelihood of each of the plurality of nodes by assigning a weight to a first likelihood of a node corresponding to a keyword not presenting in the context among a plurality of keywords corresponding to each of the plurality of nodes; calculating a third likelihood of each of all paths present in the graph based on the second likelihood of each of the plurality of nodes; and generating a summary for the context based on a path having the highest third likelihood among the paths.

Abstract: Provided are a generator and method of generating a structured query language (SQL) conversion model which converts a natural language query into a SQL query. The method includes labeling, by a training data manager, each of a plurality of keywords included in a second natural language query with a tag, the tag is a first tag means a meaningless keyword in an SQL conversion of a second tag means a meaningful keyword in the SQL conversion; converting, by a SQL converter, the second natural language query into a second SQL query by using the SQL conversion model based on the second natural language query, tags labeled for each of the plurality of keywords included in the second natural language, and a database schema; and training, by a SQL conversion model trainer, the SQL conversion model based on a loss value calculated by comparing the second SQL query and a third SQL query corresponding to the second natural language query.

Abstract: Disclosed herein is a search method performed by a server, including: receiving a user question from a user terminal; generating a user question vector for the user question; selecting similar question candidates based on a similarity to the user question vector; generating an answer to the user question based on the similar question candidates; and transmitting the answer to the user question to the user terminal.

Abstract: The present invention relates to a context-based QA generation architecture, and an object of the present invention is to generate diverse QA pairs from a single context. To achieve the object, the present invention includes a latent variable generating network including at least one encoder and an artificial neural network (Multi-Layer Perceptron: MLP) and configured to train the artificial neural network using a first context, a first question, and a first answer, and generate a second question latent variable and a second answer latent variable by applying the trained artificial neural network to a second context, an answer generating network configured to generate a second answer by decoding the second answer latent variable, and a question generating network configured to generate a second question based on a second context and the second answer.

Abstract: Aspects of the subject disclosure may include, systems and methods, for example, including receiving a user question data in a speech format or a text format, analyzing the user question data, selecting a plurality of documents from a plurality of domains corresponding to the user question data, searching for a plurality of passages including candidates for an answer value determined as being suitable for the user question data, in the plurality of documents, obtaining candidates by inputting the user question data and the plurality of passages into a plurality of MRC question and answer units, determining the answer value based on whether a reliability value of each of the candidates exceeds a threshold value, and providing the determined answer value to a user. Other embodiments are disclosed.

Abstract: The invention relates to a method and a system for improving performance of text summarization and has an object of improving performance of a technique for generating a summary from a given paragraph. According to the invention to achieve the object, a method for improving performance of text summarization includes: an a step of generating an embedding vector by vectorizing a natural language-based context; a b step of generating a graph using the embedding vector and calculating a first likelihood of each of at least one node included in the graph; a c step of generating a second likelihood by assigning a weight to the first likelihood according to a result of comparing at least one node included in the graph with the context; and a d step of calculating a third likelihood for all candidate paths present in the graph based on the second likelihood, selecting a path having a highest third likelihood, and generating a summary based on the path.

Abstract: Provided are a device and method for converting a natural language query into a structured query language (SQL) query for a database search. The method includes an operation A of labeling natural language queries included in training data, an operation B of converting the natural language queries into second SQL queries by applying the natural language queries to an SQL conversion model, an operation C of verifying the second SQL queries, and an operation D of training the SQL conversion model by comparing the second SQL queries with third SQL queries corresponding to the natural language queries of the training data.

Abstract: Provided are method and apparatus for data structuring of text. The apparatus for data structuring of text includes a processor; and a memory storing instructions executable by the processor, wherein the processor is configured to execute the instructions to: extract text and location information of the text from an image, set text units for the extracted text, assigning a first tag and a second tag to at least one of the text units, connect text units with related tags among the text units allocated the first tag and the second tag, label the connected text units as first text, second text, and third text respectively corresponding to an item name, an item value, and others based on a natural language processing model, and structure the extracted text by mapping the second text to the first text.

Abstract: The present invention relates to a method for reinforcing a multiple-choice QA model based on adversarial learning techniques, wherein incorrect answers are further generated based on a data set used in the process of training the multiple-choice QA model to enrich data which are learnable by the multiple-choice QA model.

Abstract: Provided are method and apparatus for data structuring of text. The apparatus for data structuring of text includes a data extraction unit configured to extract text and location information of the text from an image based on an optical character recognition (OCR) technique, a data processing unit configured to generate a text unit based on the text and the location information, a form classification unit configured to classify a form of the image based on the text, a labeling unit configured to label the text unit as first text, second text, and third text respectively corresponding to an item name, an item value, or others based on the classified form, a relationship identification unit configured to map and structure the second text corresponding to the first text, and a misrecognition correction unit configured to determine misrecognition of the first text and correct the first text determined to be misrecognized.

Abstract: A method of retrieving a document according to an embodiment of the present application includes: acquiring a user retrieval query; calculating a user inquiry vector in a unit of sentence from the user retrieval query and acquiring a first document candidate group based on similarity between the calculated user inquiry vector and an embedding vector of a document stored in a retrieval database; acquiring a second document candidate group based on similarity between a text included in the user retrieval query and a text of the document stored in the retrieval database; and determining a summarization target document based on the first document candidate group and the second document candidate group.

Abstract: The present invention relates to a method for reinforcing a multiple-choice QA model based on adversarial learning techniques, wherein incorrect answers are further generated based on a data set used in the process of training the multiple-choice QA model to enrich data which are learnable by the multiple-choice QA model.

Abstract: A text data structuring apparatus according to the present invention includes: a data extraction unit which extracts text included in an image and position information of the text on the basis of OCR; a data processing unit which extracts line information included in the image by using the text, the position information, and the image; a labeling unit which labels the text as keys or values; and a relationship identification unit which acquires a mapping candidate group including first text, second text, and third text labeled on the basis of the line information, calculates a first similarity score representing meaning similarity between the first text and the third text and a second similarity score representing meaning similarity between the second text and the third text, and decides text to be mapped with the third text among of the first text and the second text.

Abstract: Aspects of the subject disclosure may include, systems and methods, for example, including receiving a user question data in a speech format or a text format, analyzing the user question data, selecting a plurality of documents from a plurality of domains corresponding to the user question data, searching for a plurality of passages including candidates for an answer value determined as being suitable for the user question data, in the plurality of documents, obtaining candidates by inputting the user question data and the plurality of passages into a plurality of MRC question and answer units, determining the answer value based on whether a reliability value of each of the candidates exceeds a threshold value, and providing the determined answer value to a user. Other embodiments are disclosed.

Abstract: Provided is an artificial intelligence (AI) answering system including a user question receiver configured to receive a user question from a user terminal; a first question extender configured to generate a question template by analyzing the user question and determine whether the user question and the generated question template match; a second question extender configured to generate a similar question template by using a natural language processing and a deep learning model when the user question and the generated question template do not match; a training data builder configured to generate training data for training the second question extender by using an neural machine translation (NMT) engine; and a question answering unit configured to transmit a user question result derived through the first question extender or the second question extender to the user terminal.

Abstract: A machine reading comprehension (MRC) question and answer providing method includes receiving a user question; analyzing the user question; selecting at least one document from at least one domain corresponding to an analyzed user question and searching for a passage, which is a candidate answer determined as being suitable for the user question, in the selected at least one document; obtaining at least one correct answer candidate value by inputting the user question and a corresponding passage into each of at least one MRC question and answer unit; and determining whether the at least one correct answer candidate value is a best answer.