Abstract: A computer-implemented method is provided. The computer-implemented method includes obtaining one or more attributes of one or more entities; performing a vectorized representation on a respective attribute of the one or more attributes, the vectorized representation being encoded as a vector of d dimensions; and performing computation on vectorized representation of the one or more attributes based on attention mechanism to construct a graph neural network model. Vectors of m number of dimensions are used to represent an individual entity in the graph neural network model, (m?1) dimensions out of m dimensions represent (m?1) number of attributes, and one out of the m dimensions represents a flag bit. Constructing the graph neural network model includes performing self-attention on the individual entity, during which the flag bit is converted into a fused flag bit by fusing the individual entity's own attributes.

Abstract: The present disclosure provides a training method and apparatus for an image-text matching model, a device and a storage medium. The method includes: acquiring a positive sample and a negative sample; where the positive sample includes text and an image, the text in the positive sample is used to describe content of the image in the positive sample; the negative sample includes text and an image, the text in the negative sample describes content that is inconsistent with content of the image in the negative sample; training the image-text matching model by using the acquired positive sample and the acquired negative sample based on a manner of contrastive learning; where the image-text matching model is used to predict, for an input image and input text, whether the input text is used to describe content of the input image.

Abstract: A model training method includes: acquiring a sample set including a plurality of sample groups; the sample group includes an original image sample and original text samples; performing mask processing on the original image sample and the original text samples to generate a mask image sample and mask text samples; using the mask image sample and the mask text samples to perform adversarial training on a generator and a discriminator to obtain a target model; the generator includes a feature extraction network and an output network, the feature extraction network is used to perform feature extraction after information fusion of an input image and an input text of the generator.

Abstract: The present disclosure relates to a method for training a natural language processing model, including: obtaining a sample text of natural language; determining one or more triples in the sample text, wherein each of the triples comprises two entities in the sample text and a relation between the two entities; processing the sample text based on the triples to obtain one or more knowledge fusion vectors; and training a natural language processing model by inputting the knowledge fusion vectors into the natural language processing model to obtain a target model.

Abstract: Disclosed are an asset value evaluation method and apparatus, a model training method and apparatus, and a readable storage medium. The asset value evaluation method includes: acquiring input asset value query information for a user; when it is determined that there is historical asset interaction information of the user, determining an asset set obtained by means of making a query using the asset value query information, the asset set includes at least one asset; performing embedding representation on each asset, so as to determine an asset embedding vector of each asset, the asset embedding vector is obtained by means of training based on the relationship between each asset and an attribute, and the attribute is used for representing an inherent parameter of the asset; and inputting the asset embedding vector of each asset into a graph convolutional network model to obtain the value of each asset for the user.

Abstract: Provided in the present disclosure are a text recommendation method and apparatus, a model training method and apparatus, and a readable storage medium. The text recommendation method includes: acquiring text retrieval information from a user; when it is determined that there is historical text retrieval information for the user, determining text information of each text in a text set retrieved by using the text retrieval information; performing embedded representation on the text information of each text based on a self-attention model, and determining a text embedding vector of each text; inputting the text embedding vector of each text into a trained graph convolutional network model, to obtain the probability of interaction between the user and each text in the text set; and screening out, from the text set, target text which meets a preset interaction probability, and recommending the target text to the user.

Abstract: The present disclosure relates to a method for natural language processing, a method of training a natural language processing model, an electronic device, and a non-transitory computer-readable storage medium, and relates to the technical field of natural language processing. The method for natural language processing includes: acquiring text data; and processing the text data by using a natural language processing model to obtain output information, wherein the natural language processing model comprises a first attention model, the first attention model comprising a sequential coding matrix for adding, on the basis of the text data, sequential relation information between at least one word and other words in the text data.

Abstract: Provided in the present disclosure are a text recognition method, and a model and an electronic device, which are applied to a mode in which primary classification is first performed from different dimensions, and secondary classification is then performed, such that the meaning of text is analyzed from different dimensions, thereby improving the accuracy of text recognition. The method includes: acquiring text to be recognized, and performing primary classification on the text to obtain a plurality of text features, wherein the primary classification is used for performing feature extraction on the text from different dimensions, and there are differences between features extracted from the different dimensions (100); splicing the plurality of text features, so as to obtain spliced features (101); and performing secondary classification on the spliced features to obtain a text category corresponding to the text, wherein the secondary classification is used for classifying the spliced features (102).

Abstract: The present disclosure relates to a meter recognition method, which includes: determining embedded features of pixels in a target image of a meter, and encoding position information of the pixels to obtain encoded position features; inputting superimposed features obtained by superimposing the encoded position features and the embedded features into an encoder of a target model; wherein an input of the target model includes the labels, and an output of the target model includes coordinates of key points in a sample image of the meter. According to the present disclosure, the image of the meter can be processed by the trained target model, the coordinates of key points in the target image of the meter are outputted. It can reduce manual operations and improve efficiency, and on the other hand, it can avoid possible misoperations during manual operations, which is beneficial for improving accuracy.

Abstract: The present disclosure relates to an image recognition method and system for a display panel, a training method, and an electronic device and a non-volatile computer-readable storage medium. The image recognition method includes: acquiring an image of a display panel, wherein the image includes gate lines extending in a first direction and data lines extending in a second direction, the gate lines and the data lines intersecting to define a plurality of sub-pixel regions, and the image further includes a defect pattern; and recognizing the defect pattern in the image by using an image recognition model to obtain defect information, wherein the defect information includes at least one of a defect type or a defect position of the defect pattern, the image recognition model comprises a first attention model configured to learn a weight proportion of a feature of the defect pattern in the image.

Abstract: Provided are a method and device for recommending goods, a method and device for training a goods knowledge graph, and a method and device for training a model. The method for training a goods knowledge graph includes: constructing an initial goods knowledge graph based on a first type of triples and a second type of triples, where a format of the first type of triples is head entity-relation-tail entity, and a format of the second type of triples is entity-attribute-attribute value (S101); and training the initial goods knowledge graph based on a graph embedding model to obtain embedding vectors of entities in the trained goods knowledge graph (S102).

Abstract: A knowledge graph updating method, an apparatus, an electronic device, a storage medium and a program thereof, relates to the field of computer technology. The method includes: receiving an updating request of a technical knowledge graph, the updating request includes: a technical-event node of the technical knowledge graph; extracting historical event information corresponding to the technical-event node from the technical knowledge graph; and updating event information of the technical-event node after a current time by using a target-technical-event prediction model; wherein, the target-technical-event prediction model is obtained by training a technical-event prediction model with the historical event information.