Abstract: An apparatus for processing natural language according to an embodiment includes a collection module that collects documents having tags, a parsing module that extracts text from the collected documents and extracts tag-related information on the tag surrounding each extracted text, and a preprocessing module that generates tokens of a preset unit by tokenizing each extracted text, generates token position information for each token in full text of the document, and sets the token and the token position information as training data in matching with the tag-related information.

Abstract: The present invention relates to a golf ball location ascertaining method and a golf play information providing system. The golf ball location ascertaining method comprises a first step of ascertaining a motion location at which a predetermined motion is taken when a golfer located near a golf ball on a golf field takes the predetermined motion, and a second step of ascertaining the location of the golf ball on a golf course by using golf course data on the golf field and the motion location.

Abstract: In some embodiments, a storage device includes a nonvolatile memory device including a plurality of erase units, and a memory controller. Each of the plurality of erase units include a plurality of memory cells. The memory controller is configured to assign first memory cells of each of first erase units among the plurality of erase units as y-level cells, assign second memory cells of each of second erase units among the plurality of erase units as x-level cells, allocate a zone to at least one first erase unit among the first erase units and permit a sequential write with respect to the zone, buffer data to be written in the zone by using at least one second erase unit among the second erase units, and provide, to the external host device, information about the at least one second erase unit buffering the data to be written in the zone.

Abstract: An apparatus for training a document summarization model includes a token generation unit, a named entity recognition unit, and a model training unit. The token generation unit generates document tokens and summarization tokens. The named entity recognition unit assigns named entity token status to a summarization token, recognized as a named entity through NER, and assigns non-named entity token status to the other tokens. The model training unit obtains feature vectors by inputting the plurality of document tokens into an encoder inside a document summarization model, obtains a first loss related to the named entity token, a second loss related to the other tokens, and a total loss using a weighted value by inputting the feature vectors, the summarization tokens, the named entity token, and the non-named entity token into a decoder inside the document summarization model, and trains the document summarization model on the basis of the total loss.

Abstract: A storage controller for writing first data to a first memory cell by performing programming of the first memory cell N-times, where N is a positive integer greater than 1, includes a write amplification manager and a central processing unit. The write amplification manager checks whether the first data is invalid data before an Nth programming of the first memory cell is performed, and the central processing unit does not perform the N-th programming of the first memory cell when the first data is the invalid data.

Abstract: An apparatus for training a dialogue summary model according to an embodiment includes a parameter transferer configured to transfer one or more learning parameter values of a pre-trained natural language processing model to a sequence-to-sequence-based dialogue summary model, and a model trainer configured to train the dialogue summary model by using the transferred learning parameter values as initial values for learning parameters of each of an encoder and a decoder in the dialogue summary model.

Abstract: A storage controller for writing first data to a first memory cell by performing programming of the first memory cell N-times, where N is a positive integer greater than 1, includes a write amplification manager and a central processing unit. The write amplification manager checks whether the first data is invalid data before an Nth programming of the first memory cell is performed, and the central processing unit does not perform the N-th programming of the first memory cell when the first data is the invalid data.

Abstract: Provided are a method for generating a summary and a system therefor. The method according to some embodiments may include calculating a likelihood loss for a summary model using a first text sample and a first summary sentence corresponding to the first text sample, calculating an unlikelihood loss for the summary model using a second text sample and the first summary sentence, the second text sample being a negative sample generated from the first text sample, and updating the summary model based on the likelihood loss and the unlikelihood loss.

Abstract: A memory system which includes a flash memory including a meta block and a data block, and a memory controller including a meta data managing module managing meta data to be stored in the meta block. The meta block includes a plurality of pages. The meta data managing module compares a number of free pages of the plurality of pages with a preset number of free pages, and selects a write target page of the meta data by using one of a consecution meta write operation or a dirty meta write operation, depending on a result of the comparison.

Abstract: A sandwich panel for an automobile includes a core layer, a surface layer, and an adhesive layer. The core layer includes glass fibers and thermoplastic resins and defines an optimal weight and thickness to provide flexural performance and non-flammability. The sandwich panel provides a flame barrier layer, which is a non-combustible layer during ignition and prevents flame from leaking to the outside of the panel.

Abstract: An apparatus for embedding a sentence feature vector according to an embodiment includes a sentence acquisitor configured to acquire a first sentence and a second sentence, each including one or more words; a vector extractor configured to extract a first feature vector corresponding to the first sentence and a second feature vector corresponding to the second sentence by independently inputting each of the first sentence and the second sentence into a feature extraction network; and a vector compressor configured to compress the first feature vector and the second feature vector into a first compressed vector and a second compressed vector, respectively, by independently inputting each of the first feature vector and the second feature vector into a convolutional neural network (CNN)-based vector compression network.

Abstract: Methods and apparatuses for embedding data. The method for embedding data includes: acquiring a pretrained embedding model; generating a prompt associated with a data sample through a prompt encoder, the prompt encoder being lighter than the embedding model; generating an embedding representation of the data sample by inputting the prompt and the data sample to the embedding model; calculating a task loss by performing a predefined task by using the embedding representation; and updating the prompt encoder based on the task loss.

Abstract: The present disclosure relates to an auxiliary apparatus used to capture an image of teeth of a patient, and more particularly, to a support block that enables an image of teeth of the patient to be captured in a state where a natural dentition is maintained, and an imaging apparatus including the same. According to the dental support block of the present disclosure and the imaging apparatus including the same, imaging can be performed by reflecting a patient's normal occlusal state, thereby accurately checking and diagnosing a condition of teeth, dentition, and the like and treating accordingly.

Abstract: A storage controller for writing first data to a first memory cell by performing programming of the first memory cell N-times, where N is a positive integer greater than 1, includes a write amplification manager and a central processing unit. The write amplification manager checks whether the first data is invalid data before an Nth programming of the first memory cell is performed, and the central processing unit does not perform the N-th programming of the first memory cell when the first data is the invalid data.

Abstract: An apparatus for processing natural language according to an embodiment includes a collection module that collects documents having tags, a parsing module that extracts text from the collected documents and extracts tag-related information on the tag surrounding each extracted text, and a preprocessing module that generates tokens of a preset unit by tokenizing each extracted text, generates token position information for each token in full text of the document, and sets the token and the token position information as training data in matching with the tag-related information.

Abstract: The present invention relates to a golf ball location ascertaining method and a golf play information providing system. The golf ball location ascertaining method comprises a first step of ascertaining a motion location at which a predetermined motion is taken when a golfer located near a golf ball on a golf field takes the predetermined motion, and a second step of ascertaining the location of the golf ball on a golf course by using golf course data on the golf field and the motion location.

Abstract: A memory controller includes an interface and a control module. The interface interfaces with a memory device which includes a plurality of dies that each include a plurality of blocks. The control module groups a plurality of blocks included in different dies and manages the plurality of blocks as a super block. The control module performs scheduling to alternately perform a program on a part of an Nth super block, wherein N is a natural number, and a phased erase on an N+1st super block, and the control module completes the program on the Nth super block and the erase on the Nth super block before the program on the N+1st super block starts.

Abstract: An apparatus for training a document summarization model includes a token generation unit, a named entity recognition unit, and a model training unit. The token generation unit generates document tokens and summarization tokens. The named entity recognition unit assigns named entity token status to a summarization token, recognized as a named entity through NER, and assigns non-named entity token status to the other tokens. The model training unit obtains feature vectors by inputting the plurality of document tokens into an encoder inside a document summarization model, obtains a first loss related to the named entity token, a second loss related to the other tokens, and a total loss using a weighted value by inputting the feature vectors, the summarization tokens, the named entity token, and the non-named entity token into a decoder inside the document summarization model, and trains the document summarization model on the basis of the total loss.

Abstract: A memory controller includes an interface and a control module. The interface interfaces with a memory device which includes a plurality of dies that each include a plurality of blocks. The control module groups a plurality of blocks included in different dies and manages the plurality of blocks as a super block. The control module performs scheduling to alternately perform a program on a part of an Nth super block, wherein N is a natural number, and a phased erase on an N+1st super block, and the control module completes the program on the Nth super block and the erase on the Nth super block before the program on the N+1st super block starts.

Abstract: An apparatus for training a dialogue summary model according to an embodiment includes a parameter transferer configured to transfer one or more learning parameter values of a pre-trained natural language processing model to a sequence-to-sequence-based dialogue summary model, and a model trainer configured to train the dialogue summary model by using the transferred learning parameter values as initial values for learning parameters of each of an encoder and a decoder in the dialogue summary model.