TWM629035U - Character generator - Google Patents

Abstract

A character generation device for generating a plurality of characters, after the character generation device receives a generation request including a generation condition, encodes the generation condition into input data to be generated in a fixed format, and uses the input data to be generated according to the input data to be generated. A character generation model generates N candidate characters, and adds the N candidate characters to the candidate character string to update the candidate character string. For each scoring condition, according to the candidate character string and the scoring condition, obtain a The score of the candidate string, determine whether the total score of the at least one score is greater than a scoring threshold, when it is determined that the total score of the at least one score is greater than the scoring threshold, determine whether the candidate string meets a generation termination condition, when When it is determined that the candidate string meets the generation termination condition, the candidate string is output.

Description

text generation device

The present invention relates to a character generation device for generating characters, in particular to a character generation device for generating a plurality of characters that meet conditions.

When commercial companies generate marketing copy, in addition to understanding customers’ preferences for copy, they also need to know which words in the copy can attract customers’ attention. In the Internet age, customers are exposed to many digital channels, and the demand for copy is also higher than before. However, to produce the best marketing copy based on a large amount of copywriting, the existing method is to rely on business experts to manually generate marketing copy according to product characteristics and marketing purposes. The disadvantages of this method are: Second, the first is that the quality of the produced marketing copy will be different due to different business experts, which will jointly affect the marketing results. The second is that in order to generate the best marketing copy, it is bound to have to read a lot of copywriting and find inspiration. It can take a lot of time, so if you can provide a quick and accurate generation of the best marketing copy that will catch the customer's attention, you can reduce time and labor costs.

Therefore, the purpose of the present invention is to provide a character generation device that can instantly and automatically generate a plurality of characters that meet the conditions.

Therefore, the character generating device of the present invention is used for generating a plurality of characters, and the character generating device includes an output module, a storage module, and a processing module.

The output module is used to output the text.

The storage module is used for storing a character generation model for generating a plurality of characters, at least one scoring condition for scoring the generated characters, and a candidate character string whose initial value is empty.

The processing module is electrically connected to the output module and the storage module.

Wherein, after receiving a generation request including a generation condition, the processing module encodes the generation condition into input data to be generated in a fixed format, the generation condition includes a product category to be generated, and a Whether the product category has a preferential to-be-generated preferential attribute, the processing module uses the text generation model to generate N candidate characters according to the to-be-generated input data, and adds the N candidate characters to the candidate stored in the storage module. character string to update the candidate character string, for each scoring condition, the processing module obtains a score related to the candidate character string according to the candidate character string and the scoring condition stored in the storage module, and determines the Whether the total score of at least one score is greater than a score threshold, when the processing module determines that the total score of the at least one score is greater than the score threshold, it is determined whether the candidate string meets a generation termination condition, when the processing module determines whether the candidate string meets a generation termination condition. When it is determined that the candidate character string meets the generation termination condition, the processing module outputs the candidate character string through the output module.

The function of the present invention is: generating N candidate characters by the character generating device and adding the candidate character string, obtaining the total score according to the candidate character string using the at least one scoring condition, and determining whether the total score is greater than the score Threshold, when it is determined that it is greater than the threshold, it is determined whether the candidate character string meets the generation termination condition, and when it is determined that the candidate character string is met, the candidate character string is output, thereby using the at least one scoring condition to score the generated N candidate characters to instantly And automatically generate matching words.

Referring to FIG. 1 , an embodiment of the novel character generating device is suitable for generating multiple characters, and includes an output module 1 , a storage module 2 , and an output module 1 and the storage module 2 that are electrically connected Process module 3.

The output module 1 is used for outputting a plurality of characters.

The storage module 2 is used for storing a text generation model for generating the text, at least one scoring condition for scoring the generated text, a plurality of specific keywords, a candidate string with an empty initial value, a plurality of A training sentence related to finance, a plurality of service product categories, and a plurality of preferential keywords related to discounts. Each training sentence corresponds to a product category to be set and a preferential attribute to be set. Each service product category contains multiple product category keywords.

Referring to FIG. 1 , the text generating device may be a tablet computer, a notebook computer, a server or a personal computer, but not limited thereto.

The operation details of the text generating device will be described below through a text generating method, and sequentially include a model building program for building the text generation model, and a string generating program for generating the text.

The model building procedure includes a step 61 , a step 62 , a step 63 , a step 64 , a step 65 , a step 66 , and a step 67 .

The string generating procedure includes a step 71 , a step 72 , a step 73 , a step 74 , a step 75 , a step 76 , a step 77 , a step 78 , and a step 79 .

Referring to Figures 1 and 2, the model building procedure includes the following steps.

In step 61, for each training sentence, the processing module 3 determines whether the training sentence stored in the storage module 2 contains the product category key corresponding to one of the target product categories in the service product category any of the words. When it is determined that there is any one of the product category keywords corresponding to the target product in the training sentence, the process proceeds to step 62 . When it is determined that there is no one of the product category keywords corresponding to the target product in the training sentence, the process ends (that is, the preferential attribute setting is not continued).

In step 62, for each training sentence, the processing module 3 sets the to-be-set product category corresponding to the training sentence as the target product category.

In step 63, for each training sentence, the processing module 3 determines whether any one of the preferential keywords exists in the training sentence. When it is determined that any one of the preferential keywords exists in the training sentence, the process proceeds to step 64 . When it is determined that none of the preferential keywords exists in the training sentence, the process proceeds to step 65 .

In step 64, for each training sentence, the processing module 3 sets the to-be-set preferential attribute corresponding to the training sentence as a preferential preferential attribute.

In step 65, for each training sentence, the processing module 3 sets the to-be-set preferential attribute corresponding to the training sentence as a preferential attribute without preferential treatment.

In step 66, for each training sentence, the processing module 3 encodes the training data in a fixed format according to the training sentence, the target product category corresponding to the training sentence, and the preferential attribute corresponding to the training sentence.

For example, if one of these training sentences is to exchange foreign currency together, and the corresponding target product category and preferential attribute are foreign currency and none, then the encoded training data in a fixed format is <PRODUCT>foreign currency< DISCOUNT>No <SLOGAN> Let's exchange currency together.

In step 67, the processing module 3 uses a machine learning algorithm to establish the text generation model according to the training data. The machine learning algorithm may be a Transformer Neural Network.

Referring to FIG. 1 and FIG. 3 , the string generation process includes the following steps.

In step 71, after receiving a generation request including a generation condition, the processing module 3 encodes the generation condition into the input data to be generated in the fixed format, and the generation condition includes a product category to be generated and an instruction Find out whether the product category to be generated has preferential preferential properties to be generated. Wherein, the product category to be generated is one of the service product categories.

In step 72, the processing module 3 generates N candidate characters by using the character generation model stored in the storage module 2 according to the input data to be generated. Wherein, N can be 20, but not limited thereto.

Referring to FIG. 1 and FIG. 4 , it is worth noting that step 72 includes the following sub-steps.

In step 721, the processing module 3 uses the text generation model to generate a candidate text according to the input data to be generated.

In step 722, the processing module 3 determines whether the total number of characters in one of all candidate characters generated by executing step 721 is greater than or equal to N. When it is determined that the total number of words is not greater than or equal to N, the flow proceeds to step 723 . When it is determined that the total number of words is greater than or equal to N, the flow proceeds to step 724 .

In step 723 , the processing module 3 combines all candidate characters generated by executing step 721 after the input data to be generated as another input data to be generated, and the process returns to step 721 .

In step 724, the processing module 3 arranges all candidate characters generated by executing step 721 in order to obtain the N candidate characters.

In step 73, the processing module 3 adds the N candidate characters to the candidate character string stored in the storage module 2 to update the candidate character string.

In step 74 , for each scoring condition, the processing module 3 obtains a score related to the candidate string according to the candidate string and the scoring condition stored in the storage module 2 .

Referring to FIG. 1 and FIG. 5 , it is worth noting that the at least one scoring condition stored in the storage module 2 includes for determining whether the candidate string complies with a grammatical rule, for determining whether the candidate string exists Any one of the specific keywords and the product category to be generated for determining whether the candidate string meets the generation condition, and step 74 includes the following sub-steps.

In step 741, for the scoring condition related to the grammar rule, the processing module 3 performs part-of-speech tagging on the candidate word string to obtain a part-of-speech tagging result related to the candidate word string.

In step 742, the processing module 3 determines whether the part-of-speech tagging result complies with the grammar rule to obtain the score related to the candidate character string. Wherein, the grammatical rule may be that no connective words (for example, and, and) are allowed to end in the candidate string, but not limited thereto, and when the processing module 3 determines that the part-of-speech tagging result conforms to the grammatical rule The score obtained is, for example, 3 points, which is higher than the score obtained when it is determined that the grammar rule is not met, for example, 1 point. It is worth noting that, in other embodiments, the grammar rule can also be used to determine whether there is an associated keyword after the special keyword in the candidate string. In this case, the processing module 3 does not need to perform step 741, and It is to determine whether the candidate string has the special keyword such as interest rate, interest, discount, etc., and when it is determined that the special keyword exists, determine whether there is an association after the special keyword, such as 10% off, 8%, etc. keywords, but not limited to this; alternatively, the grammar rule can also be used to determine whether financial-related words appear adjacent to the candidate string. In this case, the processing module 3 does not need to perform step 741, but determines Whether financial-related words (for example, 888 yuan and 888 yuan) appear adjacent to each other in the candidate string, but not limited to this.

In step 743, for the scoring conditions related to the specific keywords, the processing module 3 determines whether any of the specific keywords exists in the candidate string, so as to obtain the scoring conditions related to the candidate string the rating. Wherein, the specific keywords may be words that attract customers (for example, offers or discounts), but not limited thereto, and when the processing module 3 determines that any of the specific keywords exists in the candidate string The score obtained when one of the specific keywords is determined, for example, 3 points, is higher than the score obtained when it is determined that none of the specific keywords exists, for example, 1 point. It is worth noting that, in other embodiments, the specific keywords can also be words related to company specifications (for example, the name of a competitor company or indecent words, etc.), and when the processing module 3 determines the candidate The score obtained when any of the specific keywords is present in the string, eg, 1 point, is lower than the score obtained when it is determined that any of the specific keywords is absent, eg, 3 points, But not limited to this.

In step 744, for the scoring condition related to the generation condition, the processing module 3 determines whether there is any one of the product category keywords corresponding to the product category to be generated in the candidate string, so as to determine Whether the candidate string complies with the generation condition for the product category to be generated, and then obtain the score related to the candidate string. Wherein, the score obtained when the processing module 3 determines that the candidate character string conforms to the product category to be generated is, for example, 3 points, which is higher than the score obtained when it is determined that the candidate string does not conform to the product category to be generated, such as 1 point.

In step 75, the processing module 3 determines whether the total score of the at least one score is greater than a score threshold. When the processing module 3 determines that the total score is greater than the scoring threshold, the process proceeds to step 76 . When it is determined that the total score is not greater than the score threshold, the process proceeds to step 78 .

In step 76, the processing module 3 determines whether the candidate string meets a generation termination condition. When it is determined that the candidate character string meets the generation termination condition, the flow proceeds to step 77 . When it is determined that the candidate character string does not meet the generation termination condition, the flow proceeds to step 79 . Wherein, the generation termination condition may be to determine whether there is a termination symbol (for example, a period) related to the end of sentence (EOS, End of Sentence) in the candidate string, or to determine whether the number of characters in the candidate string is equal to an output Word count threshold, but not limited to this.

In step 77 , the processing module 3 outputs the candidate string via the output module 1 .

In step 78 , the processing module 3 clears the candidate string stored in the storage module 2 , and sets the input data to be generated as the input data to be generated in step 71 , and the process returns to step 72 .

In step 79 , the processing module 3 combines the N candidate characters after the input data to be generated as another input data to be generated, and the process returns to step 72 .

To sum up, in the novel character generation device of the present invention, the processing module 3 uses the character generation model to generate N candidate characters according to the generation conditions and adds the candidate character string, and according to the The candidate string, the grammar rule, the specific keywords, and the product category to be generated for the generation condition, obtain the total score related to the candidate string, and determine whether the total score is greater than the scoring threshold, when determining When it is greater than , determine whether the candidate character string meets the generation termination condition, and when it is determined that the candidate character string is met, output the candidate character string, thereby using the at least one scoring condition to score the generated N candidate characters to instantly and automatically generate a matching Conditional words can reduce time and labor costs, so it can indeed achieve the purpose of this new model.

However, the above are only examples of the present invention, which should not limit the scope of the present invention. Any simple equivalent changes and modifications made according to the scope of the patent application for this new model and the contents of the patent specification are still within the scope of the present invention. within the scope of this new patent.

1: output module 2: storage module 3: Processing modules 61~67: Steps 71~79: Steps 721~724: Steps 741~744: Steps

Other features and effects of the present invention will be clearly presented in the embodiments with reference to the drawings, wherein: FIG. 1 is a block diagram illustrating an embodiment of the novel character generation device of the present invention; FIG. 2 is a flow chart illustrating a model building procedure for realizing a character generation method using an embodiment of the novel character generation device; FIG. 3 is a flow chart illustrating a string generation program for realizing the character generation method using an embodiment of the novel character generation device; FIG. 4 is a flow chart illustrating how the text generating device generates text character by character; and FIG. 5 is a flow chart illustrating how the text generating apparatus obtains the score.

1: output module

2: storage module

3: Processing modules

Claims (8)

A character generation device for generating a plurality of characters, comprising: an output module for outputting the characters; a storage module for storing a character generation model for generating a plurality of characters, at least one for Scoring conditions for the text generated by scoring, and a candidate string whose initial value is empty; and a processing module electrically connected to the output module and the storage module; wherein, the processing module receives a After the conditional generation request, encode the generation condition into a fixed-format input data to be generated, the generation condition includes a product category to be generated, and a to-be-generated discount attribute indicating whether the to-be-generated product category has a discount, the The processing module uses the text generation model to generate N candidate characters according to the input data to be generated, and adds the N candidate characters to the candidate character string stored in the storage module to update the candidate character string. a scoring condition, the processing module obtains a score related to the candidate string according to the candidate string and the scoring condition stored in the storage module, and determines whether the total score of the at least one score is greater than a scoring threshold , when the processing module determines that the total score of the at least one score is greater than the scoring threshold, determines whether the candidate character string meets a generation termination condition, and when the processing module determines that the candidate character string meets the generation termination condition, The processing module outputs the candidate string via the output module. The character generation device of claim 1, wherein when the processing module determines that the candidate character string does not meet the generation termination condition, the N candidate characters are combined after the input data to be generated as another Input data to be generated, and repeatedly use the text to generate according to the other input data to be generated generating another N candidate characters into a model, and updating the candidate character string and obtaining at least one other score related to the candidate character string, and determining whether the total score of the at least one other score is greater than the score threshold, and determining the Whether the candidate string meets the generation termination condition, until the candidate string meets the generation termination condition. The character generation device according to claim 1, wherein the processing module generates a candidate character by using the character generation model stored in the storage module according to the input data to be generated, and the processing module determines all characters using the character generation model Whether the total number of characters in one of the generated candidate characters is greater than or equal to N, when it is determined that the total number of characters is not greater than or equal to N, all candidate characters generated by using the character generation model are combined with the input data to be generated. As another input data to be generated, and according to the further input data to be generated, use the text generation model to generate another candidate text repeatedly, and determine whether the total number of characters of all the other candidate texts generated by using the text generation model is not greater than or equal to N, until the total number of characters is greater than or equal to N, when the processing module determines that the total number of characters is greater than or equal to N, the character generation device arranges all candidate characters generated by the character generation model in order to obtain the N candidate words. The text generation device of claim 1, wherein one of the at least one scoring condition stored in the storage module is used to determine whether the candidate character string complies with a grammar rule, and for the Scoring condition, the processing module determines whether the candidate character string complies with the grammar rule to obtain the score related to the candidate character string. The character generation device according to claim 1, wherein the storage module further stores There are a plurality of specific keywords, and one of the at least one scoring condition stored in the storage module is used to determine whether any one of the specific keywords exists in the candidate string. The scoring condition of the word, the processing module is to determine whether any one of the specific keywords exists in the candidate word string, so as to obtain the score related to the candidate word string. The text generation device according to claim 1, wherein the storage module further stores a plurality of service product categories, the to-be-generated product category is one of the service product categories, and each service product category includes a plurality of service product categories Product category keyword, one of the at least one scoring condition stored in the storage module is a product category to be generated for determining whether the candidate string meets the generating condition, and for the scoring condition related to the generating condition, The processing module determines whether there is any one of the product category keywords corresponding to the product category to be generated in the candidate string, so as to determine whether the candidate string meets the generation condition for the to-be-generated product category, and then obtains The score relative to the candidate string. The text generation device of claim 1, wherein when the processing module determines that the total score of the at least one score is not greater than the score threshold, the processing module clears the candidate string stored in the storage module, The input data to be generated is set as the input data to be generated after the processing module encodes the generation conditions, and the text generation model is repeatedly used to generate another N candidate characters according to the input data to be generated, and the candidate characters are updated. character string and obtain at least one other score related to the candidate character string, and determine whether the total score of the at least one other score is greater than the score threshold, and determine whether the candidate character string meets the generation termination condition, until the candidate character The string satisfies this generation termination condition. The text generation device according to claim 1, wherein the storage module further stores a plurality of training sentences related to finance, a plurality of service product categories, and a plurality of preferential keywords related to discounts, each training sentence Correspondingly, there is a product category to be set, and a preferential attribute to be set. Each service product category includes a plurality of product category keywords. For each training sentence, the processing module determines whether there is a corresponding service product category in the training sentence. any one of the product category keywords under one of the target product categories, for each training sentence, when the processing module determines that the training sentence contains any one of the product category keywords under the target product category In the case of the above, the processing module sets the to-be-set product category corresponding to the training sentence as the target product category, and for each training sentence, the processing module determines whether there is any task for the preferential keywords in the training sentence. One, for each training sentence, when the processing module determines that there is any one of the preferential keywords in the training sentence, the processing module sets the to-be-set preferential attribute corresponding to the training sentence to have The preferential attribute of the preferential treatment. For each training sentence, when the processing module determines that there is no one of the preferential keywords in the training sentence, the processing module corresponds to the preferential attribute to be set corresponding to the training sentence. It is set as a preferential attribute without preferential treatment. For each training sentence, the processing module encodes the training data in a fixed format according to the training sentence, the target product category corresponding to the training sentence, and the preferential attribute corresponding to the training sentence. And according to the training data, a machine learning algorithm is used to establish the text generation model.

Images