Text recognition method and device, computer-readable storage medium and electronic device

Abstract

The present application relates to the technical field of text recognition, and specifically to a text recognition method and a text recognition device, as well as a computer-readable storage medium and an electronic device, which solve the problems of inaccurate and inefficient text recognition. The text recognition method, by making adjacent character strings in multiple character strings partially overlap, can make multiple character strings reflect the relationship between the contexts of the text to be recognized. Then, multiple character strings are converted into word vectors to obtain multiple word vectors, and word vector recognition results corresponding to each of the multiple word vectors are generated based on the multiple word vectors to determine whether the text corresponding to the word vector is a functional text or a non-functional text, and the text recognition result of the text to be recognized is determined by combining the multiple word vector recognition results, and the relationship between the contexts of the text to be recognized is recognized, thereby improving the accuracy of text recognition. In addition, the text recognition method of the present application does not require manual indexing, which improves the efficiency of text recognition.

Description

Technical Field

[0001] The present application relates to the field of text processing technology, and in particular to a text recognition method and device, a computer-readable storage medium, and an electronic device. Background Art

[0002] Patent documents mainly include three parts: technical problems, technical solutions and technical effects. Patent documents can be classified in detail according to the technical effects part.

[0003] At present, there is a method of using the bibliographic information of patents to identify functional texts, thereby classifying patent documents. However, the bibliographic information of patents is too broad to accurately identify functional texts, and it is impossible to classify patent documents in detail. In the prior art, in order to improve the accuracy of identifying functional texts, functional texts are mainly determined by rule-based indexing or manual indexing. Rule-based indexing determines functional texts by identifying specific grammatical patterns, which can easily miss other expressions that cannot be covered by the grammatical patterns, and thus may miss important patent information, resulting in low accuracy in text recognition. Although manual indexing is more accurate, it requires a lot of manpower, resulting in low efficiency in text recognition. Summary of the invention

[0004] In view of this, embodiments of the present application provide a text recognition method and device, a computer-readable storage medium, and an electronic device, which solve the problems of inaccurate and low efficiency in text recognition.

[0005] In a first aspect, an embodiment of the present application provides a text recognition method, comprising: determining multiple character strings corresponding to the text to be recognized based on the text to be recognized, wherein adjacent character strings in the multiple character strings partially overlap; performing word vector conversion on the multiple character strings to obtain multiple word vectors, wherein the multiple word vectors are in a one-to-one correspondence with the multiple character strings; generating multiple word vector recognition results corresponding to each of the multiple word vectors based on the multiple word vectors, wherein the word vector recognition results are functional text or non-functional text; and determining the text recognition result of the text to be recognized based on the word vector recognition results corresponding to each of the multiple word vectors.

[0006] In combination with the first aspect of the present application, in some embodiments, the number of multiple character strings is M, M is a positive integer greater than 1, and multiple character strings corresponding to the text to be recognized are determined based on the text to be recognized, including: determining the first character string of a preset character string length based on the text to be recognized; taking the characters in the Nth character string as the starting characters of the N+1th character string, decomposing the text to be recognized based on the preset character string length and the preset decomposition step size, and obtaining the N+1th character string, wherein N is a positive integer greater than or equal to 1 and less than M.

[0007] In combination with the first aspect of the present application, in some embodiments, the text recognition result of the text to be recognized is determined based on the word vector recognition results corresponding to each of the multiple word vectors, including: using a voting mechanism to perform a voting operation on the word vector recognition results corresponding to each of the multiple word vectors to determine the text recognition result of the text to be recognized.

[0008] In combination with the first aspect of the present application, in some embodiments, a voting mechanism is adopted to perform a voting operation on the word vector recognition results corresponding to each of the multiple word vectors to determine the text recognition result of the text to be recognized, including: determining multiple units to be recognized corresponding to the text to be recognized based on the text to be recognized, wherein the unit to be recognized corresponds to at least one word vector; for each of the multiple units to be recognized, a voting mechanism is adopted to perform a voting operation on the word vector recognition results corresponding to the unit to be recognized to determine the text recognition result of the unit to be recognized; based on the text recognition results corresponding to each of the multiple units to be recognized, determine the text recognition result corresponding to the text to be recognized.

[0009] In combination with the first aspect of the present application, in some embodiments, the text to be identified is a patent text, and the unit to be identified includes at least one of a sentence, a paragraph, and a text module in the patent text; wherein the text module includes at least one of an abstract module, a claim module, and a specification module.

[0010] In combination with the first aspect of the present application, in some embodiments, for each unit to be recognized among multiple units to be recognized, a voting mechanism is adopted to perform a voting operation on the word vector recognition result corresponding to the unit to be recognized to determine the text recognition result of the unit to be recognized, including: for each unit to be recognized among the multiple units to be recognized, if in the word vector recognition result corresponding to the unit to be recognized, the number of effective texts is greater than or equal to the number of non-effective texts, determining that the text recognition result of the unit to be recognized is effective text.

[0011] In combination with the first aspect of the present application, in some embodiments, multiple word vector recognition results corresponding to each word vector are generated based on multiple word vectors, including: using an efficacy recognition model to generate multiple word vector recognition results corresponding to each word vector based on multiple word vectors, wherein the efficacy recognition model is used to generate a word vector recognition result corresponding to an input word vector based on the input word vector.

[0012] In combination with the first aspect of the present application, in some embodiments, before using the efficacy recognition model to generate multiple word vector recognition results corresponding to each word vector based on multiple word vectors, it also includes: determining a training text and a text recognition result corresponding to the training text; based on the training text and the text recognition result corresponding to the training text, determining multiple word vector samples corresponding to the training text and the word vector recognition results corresponding to the multiple word vector samples, establishing an initial network model, and training the initial network model based on the multiple word vector samples and the word vector recognition results corresponding to the multiple word vector samples to generate an efficacy recognition model.

[0013] In combination with the first aspect of the present application, in some embodiments, the training text includes a first language training text and a second language training text, the first language training text includes text content written in the first language, the second language training text includes text content written in the second language, the text recognition result corresponding to the first language training text is a first text recognition result, and the text recognition result corresponding to the second language training text is a second text recognition result; determining the training text and the text recognition result corresponding to the training text includes: acquiring the first language training text and the second language training text; based on the first language training text and the first text recognition result, obtaining a first function marked text corresponding to the first language training text; based on the first function marked text, determining the second text recognition result corresponding to the second language training text.

[0014] In combination with the first aspect of the present application, in some embodiments, based on the first function-marked text, determining the second text recognition result corresponding to the second language training text includes: translating the first function-marked text to obtain a translated text corresponding to the first function-marked text, wherein the translated text is expressed in the second language; and using a similarity algorithm to determine the second text recognition result based on the translated text corresponding to the first function-marked text.

[0015] In combination with the first aspect of the present application, in some embodiments, the training text is a patent text, and obtaining the first language training text and the second language training text includes: obtaining patent family text data including multiple different languages; screening the first language training text and the second language training text based on the patent family text data, wherein the first language training text includes efficacy identification paragraph information, and the second language training text includes the patent text to be efficacy marked.

[0016] In conjunction with the first aspect of the present application, in some embodiments, the first language includes Japanese, and the first language training text includes Japanese patent text.

[0017] In a second aspect, a text recognition device provided by an embodiment of the present application includes: a splitting module, configured to determine multiple character strings corresponding to the text to be recognized based on the text to be recognized, wherein adjacent character strings in the multiple character strings partially overlap; a conversion module, configured to perform word vector conversion on the multiple character strings to obtain multiple word vectors, wherein the multiple word vectors are in a one-to-one correspondence with the multiple character strings; a generation module, configured to generate multiple word vector recognition results corresponding to each of the multiple word vectors based on the multiple word vectors, wherein the word vector recognition results are functional text or non-functional text; and a determination module, configured to determine the text recognition result of the text to be recognized based on the word vector recognition results corresponding to each of the multiple word vectors.

[0018] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, which stores instructions. When the instructions are executed by a processor of an electronic device, the electronic device can execute the text recognition method of any of the above embodiments.

[0019] In a fourth aspect, an embodiment of the present application provides an electronic device, the electronic device comprising: a processor; a memory for storing computer-executable instructions; and a processor for executing computer-executable instructions to implement the text recognition method of any of the above embodiments.

[0020] The text recognition method provided by the embodiment of the present application determines multiple character strings corresponding to the text to be recognized based on the text to be recognized, and the adjacent character strings in the multiple character strings are partially overlapped, so that the multiple character strings can reflect the relationship between the contexts of the text to be recognized. Then, the multiple character strings are converted into word vectors to obtain multiple word vectors, and the word vector recognition results corresponding to each of the multiple word vectors are generated based on the multiple word vectors to determine whether the text corresponding to the word vector is a functional text or a non-functional text, and the text recognition result of the text to be recognized is determined according to the word vector recognition results corresponding to each of the multiple word vectors. The text recognition can be performed more meticulously, and the text recognition result of the text to be recognized is determined by combining the multiple word vector recognition results, and the relationship between the contexts of the text to be recognized is recognized, which improves the accuracy of text recognition. In addition, the text recognition method of the present application does not require manual indexing, which improves the efficiency of text recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 The figure is a schematic diagram of an application scenario of a text recognition method provided in an embodiment of the present application.

[0022] Figure 2 The figure is a flowchart of a text recognition method provided in one embodiment of the present application.

[0023] Figure 3 Shown is a flowchart of a text recognition method provided in another embodiment of the present application.

[0024] Figure 4a FIG. 1 is a schematic diagram of a method for determining a character string provided by an embodiment of the present application.

[0025] Figure 4b FIG. 1 is a schematic diagram of a method for determining a character string provided in another embodiment of the present application.

[0026] Figure 5 Shown is a flowchart of a text recognition method provided in another embodiment of the present application.

[0027] Figure 6a Shown is a flowchart of a text recognition method provided in another embodiment of the present application.

[0028] Figure 6b The figure is a schematic diagram of a correspondence between a unit to be identified and a character string provided in an embodiment of the present application.

[0029] Figure 6c Shown is a schematic diagram of a correspondence between a unit to be identified and a character string provided in another embodiment of the present application.

[0030] Figure 7 Shown is a flowchart of a text recognition method provided in another embodiment of the present application.

[0031] Figure 8 Shown is a flowchart of a text recognition method provided in another embodiment of the present application.

[0032] Fig. 9 Shown is a flowchart of a text recognition method provided in another embodiment of the present application.

[0033] Fig.10 Shown is a flowchart of a text recognition method provided in another embodiment of the present application.

[0034] Fig.11a Shown is a flowchart of a text recognition method provided in another embodiment of the present application.

[0035] Fig.11b Shown is a schematic diagram of a first efficacy marking text provided in an embodiment of the present application.

[0036] Fig.12 Shown is a flowchart of a text recognition method provided in another embodiment of the present application.

[0037] Fig.13 Shown is a flowchart of a text recognition method provided in another embodiment of the present application.

[0038] Fig.14Shown is a structural schematic diagram of a text recognition device provided in one embodiment of the present application.

[0039] Fig.15 FIG. 1 is a schematic diagram of the structure of a character string determination unit provided in one embodiment of the present application.

[0040] Fig.16 Shown is a structural schematic diagram of a determination module provided in one embodiment of the present application.

[0041] Fig.17 Shown is a schematic diagram of the structure of a voting unit provided in one embodiment of the present application.

[0042] Fig.18 Shown is a structural schematic diagram of a voting operation subunit provided in an embodiment of the present application.

[0043] Fig.19 Shown is a schematic diagram of the structure of a generation module provided in one embodiment of the present application.

[0044] Fig. 20 Shown is a structural schematic diagram of a text recognition device provided in another embodiment of the present application.

[0045] Fig.21 Shown is a schematic structural diagram of a sample preprocessing module provided in one embodiment of the present application.

[0046] Fig. 22 FIG. 1 is a schematic diagram of the structure of a second sample preprocessing unit provided in one embodiment of the present application.

[0047] Fig.23 Shown is a schematic structural diagram of a sample preprocessing module provided in another embodiment of the present application.

[0048] Fig.24 FIG. 4 is a schematic structural diagram of a first sample preprocessing unit provided in another embodiment of the present application.

[0049] Fig.25 Shown is a schematic structural diagram of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0051] The technical solution provided in the present application can be applied in smart terminals (such as tablet computers, mobile phones, etc.) to enable the smart terminals to have relevant functions, such as the recognition function of functional texts, the translation function of texts, etc.

[0052] Exemplarily, the technical solution provided by the present application can be applied in patent search scenarios. Specifically, in patent search scenarios, especially in patent efficacy search scenarios, the technical solution provided by the present application can be used to judge the subject matter to be expressed by the patent document through technical efficacy, and the patents can be classified accordingly to improve the efficiency of patent search. Optionally, the present application uses a deep learning network model to identify technical efficacy paragraphs, which can narrow the scope of searching for technical efficacy words. Optionally, the present application can directly present the technical efficacy content of the patent, which can save users time in reading patent documents and help users quickly understand the subject matter and technical features of the patent.

[0053] In addition to the patent search scenarios mentioned above, this application can also be applied to many other scenarios in the form of a server. Figure 1 Give a brief introduction.

[0054] Figure 1 The figure is a schematic diagram of an application scenario of a text recognition method provided in an embodiment of the present application. Figure 1 The illustrated scenario includes a server 110 and a client 120 in communication with the server 110. Specifically, the server 110 is used to determine a plurality of word vectors based on the text to be recognized, the plurality of word vectors being used to represent semantic and grammatical information of the text to be recognized; generate a plurality of word vector recognition results corresponding to each of the word vectors based on the plurality of word vectors, the word vector recognition results being functional text or non-functional text; and determine a text recognition result of the text to be recognized based on the plurality of word vector recognition results corresponding to each of the word vectors.

[0055] For example, in actual application, the client 120 can receive the text to be recognized input by the user, and send the received text to be recognized to the server 110. The server 110 generates a text recognition result based on the received text to be recognized, and sends the generated text recognition result to the client 120. The client 120 presents the received text recognition result to the user.

[0056] Exemplary Methods

[0057] Combine the following Figures 2 to 13 A brief introduction to the text recognition method provided in this application is given.

[0058] Figure 2 FIG. 1 is a flow chart of a text recognition method provided by an embodiment of the present application. Figure 2 As shown, the text recognition method provided in the embodiment of the present application includes the following steps.

[0059] Step 210: determining multiple character strings corresponding to the text to be recognized based on the text to be recognized.

[0060] Exemplarily, adjacent character strings in multiple character strings partially overlap. For example, the text to be recognized is: "The utility model relates to the field of agricultural machinery technology, and specifically to a structure of the overall configuration of a protective cover of the threshing part of a crawler harvester. It includes a threshing frame and an overall protective cover." The multiple character strings corresponding to the text to be recognized may be: "Span1: The utility model relates to the field of agricultural machinery technology", "Span2: The field of machinery technology, specifically to a protection of the threshing part of a crawler harvester", "Span3: The structure of the overall configuration of the protective cover of the threshing part. It includes a threshing frame", "Span4: Structure. It includes a threshing frame and an overall protective cover." "Span" represents a character string, "Span1" represents the first character string, "Span2" represents the second character string, and so on. Each character string may include the same number of characters or a different number of characters, which is not specifically limited in this application. Adjacent character strings may include partially identical characters. For example, both the first character string and the second character string include the following characters "field of machinery technology".

[0061] The text to be recognized can be a sentence, a paragraph, or a text, and this application does not make any specific limitation.

[0062] Step 220, convert the multiple character strings into word vectors to obtain multiple word vectors.

[0063] Exemplarily, multiple word vectors are in one-to-one correspondence with multiple strings. Word2vec (word to vector) can be used to convert multiple strings into word vectors. Word2vec is a model used to generate word vectors. Input a string into Word2vec to get the corresponding word vector.

[0064] Step 230, generating a plurality of word vector recognition results corresponding to each of the word vectors based on the plurality of word vectors.

[0065] Exemplarily, the word vector recognition result is an effective text or a non-effective text. Each word vector corresponds to a word vector recognition result. The effective text is the text that represents the technical effect, and the non-effective text is the text other than the text that represents the technical effect.

[0066] In one embodiment of the present application, the text to be identified may be a patent text, the efficacy text is the text in the patent text that represents the technical effect, and the non-efficacy text is the other text in the patent text excluding the text that represents the technical effect.

[0067] Step 240, determining a text recognition result of the text to be recognized based on the word vector recognition results corresponding to each of the multiple word vectors.

[0068] Specifically, if the text to be recognized is a sentence, the text recognition result of the text to be recognized may be whether the sentence is functional text or non-functional text. If the text to be recognized is a paragraph, the text recognition result of the text to be recognized may be whether the paragraph is functional text or non-functional text. If the text to be recognized is a text, the text recognition result of the text to be recognized may be that a sentence or a paragraph in the text is marked as functional text.

[0069] By determining multiple character strings corresponding to the text to be recognized and converting the multiple character strings into word vectors to obtain multiple word vectors, multiple word vectors can be used to comprehensively represent the text to be recognized, providing accurate data support for text recognition.

[0070] The text recognition method provided by the embodiment of the present application determines multiple character strings corresponding to the text to be recognized based on the text to be recognized, and the adjacent character strings in the multiple character strings are partially overlapped, so that the multiple character strings can reflect the relationship between the contexts of the text to be recognized. Then, the multiple character strings are converted into word vectors to obtain multiple word vectors, and the word vector recognition results corresponding to each of the multiple word vectors are generated based on the multiple word vectors to determine whether the text corresponding to the word vector is a functional text or a non-functional text, and the text recognition result of the text to be recognized is determined according to the word vector recognition results corresponding to each of the multiple word vectors. The text recognition can be performed more meticulously, and the text recognition result of the text to be recognized is determined by combining the multiple word vector recognition results, and the relationship between the contexts of the text to be recognized is recognized, which improves the accuracy of text recognition. In addition, the text recognition method of the present application does not require manual indexing, which improves the efficiency of text recognition.

[0071] Figure 3 FIG. 1 is a flowchart of a text recognition method provided by another embodiment of the present application. Figure 2 The present application is extended based on the embodiment shown Figure 3 The embodiment shown is described below in detail. Figure 3 The embodiment shown and Figure 2 The differences between the embodiments shown are shown, and the similarities are not repeated. Figure 3 As shown, the step of determining multiple character strings corresponding to the text to be recognized based on the text to be recognized includes the following steps.

[0072] Step 310: Determine a first character string of a preset character string length based on the text to be recognized.

[0073] Exemplarily, the number of the multiple character strings is M, where M is a positive integer greater than or equal to 1. The preset character string length may be a preset character string length, for example, 256 characters or 128 characters, and the preset character string length may be selected according to actual needs, and this application does not make any specific limitation.

[0074] Step 320 , using the characters in the Nth character string as the starting characters of the N+1th character string, decomposing the text to be recognized based on a preset character string length and a preset decomposition step length, to obtain the N+1th character string.

[0075] Exemplarily, N is a positive integer greater than or equal to 1 and less than M. The preset decomposition step size may be 128 characters or 64 characters. The preset decomposition step size may be selected according to actual needs and is not specifically limited in this application.

[0076] Exemplarily, a sliding window method may be used to determine multiple character strings. Figure 4a As shown, the text to be recognized is: "The utility model relates to the technical field of agricultural machinery, and specifically to a structure of the overall configuration of the protective cover of the threshing part of a crawler harvester. Including a threshing frame, and also including an overall protective cover.". The preset string length of the sliding window can be 20 characters, and the preset decomposition step length of the sliding window can be 10 characters. The multiple strings corresponding to the text to be recognized are: "Span1: The utility model relates to agricultural machinery", "Span2: Relating to the technical field of agricultural machinery", "Span3: The technical field of machinery, specifically relates to", "Span4:, specifically relates to a crawler", "Span5: A crawler harvester threshing", "Span6: The protective cover of the threshing part of the harvester", "Span7: The structure of the overall configuration of the protective cover", "Span8: The structure of the overall configuration of the protective cover. Including threshing", "Span9: Structure. Including threshing frame, also", "Span10: Threshing frame, also including an overall protective cover", "Span11: Including an overall protective cover. Padding Padding". That is, Figure 4a In the example, the number of strings M is 11. If N is 1, the first string is Span1, and the N+1th string is Span2, that is, a character in Span1 is the starting character of Span2. "Padding" is an automatic placeholder. If the length of the last string is less than the preset string length, the automatic placeholder can be used to supplement the length of the string so that the length of the last string is equal to the preset string length. Figure 4b As shown, Figure 4b A rectangle in the figure represents a character string, and the rectangles with hatched lines represent the overlapping parts of adjacent character strings.

[0077] By first determining the first character string of a preset character string length, and then taking the characters in the Nth character string as the starting characters of the N+1th character string, and decomposing the text to be recognized based on the preset character string length and the preset decomposition step, the N+1th character string is obtained. This allows each character string to include the same number of characters, and adjacent character strings to include the same number of overlapping characters, thereby improving the uniformity of character string decomposition, better representing the semantic and grammatical relationship between character strings, and thus providing more accurate data support for text recognition.

[0078] Figure 5 FIG. 1 is a flowchart of a text recognition method provided by another embodiment of the present application. Figure 2 The present application is extended based on the embodiment shown Figure 5 The embodiment shown is described below in detail. Figure 5 The embodiment shown and Figure 2 The differences between the embodiments shown are shown, and the similarities are not repeated. Figure 5 As shown, the step of determining the text recognition result of the text to be recognized based on the word vector recognition results corresponding to each of the multiple word vectors includes the following steps.

[0079] Step 510, adopting a voting mechanism, voting on the word vector recognition results corresponding to each of the multiple word vectors to determine the text recognition result of the text to be recognized.

[0080] Exemplarily, for a text to be recognized, if in the word vector recognition result corresponding to the text to be recognized, the number of effective texts is greater than or equal to the number of non-effective texts, the text recognition result of the text to be recognized is determined to be effective text. If in the word vector recognition result corresponding to the text to be recognized, the number of effective texts is less than the number of non-effective texts, the text recognition result of the unit to be recognized is determined to be non-effective text.

[0081] Since adjacent character strings in multiple character strings partially overlap, a voting mechanism is adopted to vote on the word vector recognition results corresponding to each of the multiple word vectors to determine the text recognition result of the text to be recognized. This can identify the relationship between the contexts of the text to be recognized, thereby improving the accuracy of text recognition. At the same time, the voting mechanism complies with the voting principle of minority obeys majority, further improving the accuracy of text recognition.

[0082] Figure 6a FIG. 1 is a flowchart of a text recognition method provided by another embodiment of the present application. Figure 5 The present application is extended based on the embodiment shown Figure 6a The embodiment shown is described below in detail. Figure 6a The embodiment shown and Figure 5 The differences between the embodiments shown are shown, and the similarities are not repeated. Figure 6aAs shown, a voting mechanism is adopted to vote on the word vector recognition results corresponding to each of the multiple word vectors to determine the text recognition result of the text to be recognized, including the following steps.

[0083] Step 610: Determine a plurality of units to be recognized corresponding to the text to be recognized based on the text to be recognized.

[0084] Exemplarily, the unit to be recognized corresponds to at least one word vector. If the text to be recognized is a text, the unit to be recognized can be a sentence or a paragraph in the text. The character string can be a text segment obtained by decomposing the sentence or paragraph. The word vector corresponds to the character string one by one.

[0085] In one embodiment of the present application, the text to be recognized is a patent text, and the unit to be recognized can be at least one of a sentence, a paragraph, and a text module in the patent text. The text module can be at least one of an abstract module, a claim module, and a specification module.

[0086] Step 620, for each unit to be recognized among the multiple units to be recognized, a voting mechanism is used to vote on the word vector recognition result corresponding to the unit to be recognized, so as to determine the text recognition result of the unit to be recognized.

[0087] Exemplarily, the text recognition result of the unit to be recognized may be functional text or non-functional text.

[0088] Specifically, Figure 6b and 6c As shown in , T is a unit to be identified. There are four possible correspondences between a unit to be identified and a string. The first one is, Figure 6b As shown in , a string contains a unit to be recognized. The second type is that there is no corresponding relationship between a string and a unit to be recognized, such as Figure 6c As shown, the relationship between Span1 and Span7 and T belongs to the second case. The third case is that a string overlaps with a unit to be identified or a unit to be identified contains at least one string, such as Figure 6cAs shown, the relationship between Span2 to Span6 and T belongs to the third case. For the first case, the text recognition result of the unit to be recognized depends on the word vector recognition result corresponding to the string, that is, if the word vector recognition result corresponding to the string is an effective text, then the text recognition result of the unit to be recognized is an effective text, if the word vector recognition result corresponding to the string is a non-effective text, then the text recognition result of the unit to be recognized is a non-effective text. For the second case, there is no correlation between the text recognition result of the unit to be recognized and the word vector recognition result corresponding to the string. For the third case, the text recognition result of the unit to be recognized depends on the word vector recognition result corresponding to the part that overlaps with a unit to be recognized and the string contained in the unit to be recognized. That is, the text recognition result of the unit to be recognized T depends on the word vector recognition results of Span2 to Span6. In the third case, a voting mechanism is needed to vote on the word vector recognition results corresponding to the unit to be recognized to determine the text recognition result of the unit to be recognized.

[0089] Step 630: Determine the text recognition result corresponding to the text to be recognized based on the text recognition results corresponding to each of the multiple units to be recognized.

[0090] Exemplarily, the text recognition result corresponding to the text to be recognized may be a partial sentence or paragraph marked as functional text.

[0091] By determining the text recognition result of the unit to be recognized, and determining the text recognition result corresponding to the text to be recognized based on the text recognition results corresponding to each of the multiple units to be recognized, it is possible to identify whether multiple parts in a text are functional texts, thereby marking the parts belonging to the functional text for user viewing convenience.

[0092] Figure 7 FIG. 6 is a flowchart of a text recognition method provided by another embodiment of the present application. Figure 7 The embodiment shown is described below in detail. Figure 7 The differences between the embodiment shown in FIG. 6 and the similarities are not repeated here. Figure 7 As shown, for each unit to be recognized among multiple units to be recognized, a voting mechanism is adopted to vote on the word vector recognition result corresponding to the unit to be recognized, and the steps of determining the text recognition result of the unit to be recognized include the following steps.

[0093] Step 710, for each unit to be recognized among the multiple units to be recognized, if the number of functional texts in the word vector recognition result corresponding to the unit to be recognized is greater than or equal to the number of non-functional texts, determine that the text recognition result of the unit to be recognized is functional text.

[0094] Step 720, for each unit to be recognized among the multiple units to be recognized, if the number of effective texts in the word vector recognition result corresponding to the unit to be recognized is less than the number of non-effective texts, determine that the text recognition result of the unit to be recognized is non-effective text.

[0095] Specifically, for Figure 6c In the third case shown, the text recognition result of the unit to be recognized is the comparison result of the number of effective texts and the number of non-effective texts in the word vector recognition results of Span2 to Span6. For example, the word vector recognition result of Span2 is effective text, the word vector recognition result of Span3 is effective text, the word vector recognition result of Span4 is effective text, the word vector recognition result of Span5 is non-effective text, and the word vector recognition result of Span6 is non-effective text, that is, the number of effective texts is 3 and the number of non-effective texts is 2, then the number of effective texts is greater than the number of non-effective texts, and the text recognition result of the unit to be recognized is effective text.

[0096] For each unit to be recognized among the multiple units to be recognized, the voting principle of minority obeys majority is adopted to determine the text recognition result of the unit to be recognized, thereby further improving the accuracy of text recognition.

[0097] Figure 8 FIG. 1 is a flowchart of a text recognition method provided by another embodiment of the present application. Figure 2 The present application is extended based on the embodiment shown Figure 8 The embodiment shown is described below in detail. Figure 8 The embodiment shown and Figure 2 The differences between the embodiments shown are shown, and the similarities are not repeated. Figure 8 As shown, the steps of generating word vector recognition results corresponding to multiple word vectors respectively based on multiple word vectors include the following steps.

[0098] Step 810, using the efficacy recognition model, generates word vector recognition results corresponding to multiple word vectors based on multiple word vectors.

[0099] Exemplarily, the efficacy recognition model is used to generate a word vector recognition result corresponding to the input word vector based on the input word vector.

[0100] By utilizing the efficacy recognition model to generate word vector recognition results corresponding to multiple word vectors, there is no need for manual indexing and the word vector recognition results can be automatically obtained, thus reducing labor costs.

[0101] Fig. 9 FIG. 1 is a flowchart of a text recognition method provided by another embodiment of the present application. Figure 8 The present application is extended based on the embodiment shown Fig. 9 The embodiment shown is described below in detail. Fig. 9 The embodiment shown and Figure 8 The differences between the embodiments shown are shown, and the similarities are not repeated. Fig. 9 As shown, before the step of using the efficacy recognition model to generate multiple word vector recognition results corresponding to each word vector based on multiple word vectors, the following steps are also included.

[0102] Step 910: Determine the training text and the text recognition result corresponding to the training text.

[0103] Specifically, the training text mentioned in step 910 corresponds to the text to be recognized in the above embodiment.

[0104] Step 920, based on the training text and the text recognition result corresponding to the training text, determine a plurality of word vector samples corresponding to the training text and the word vector recognition results corresponding to each of the plurality of word vector samples.

[0105] Specifically, the method for determining the multiple word vector samples corresponding to the training text refers to the method for determining the multiple word vectors corresponding to the text to be recognized in the above embodiment, which will not be repeated here. Determining the multiple word vector recognition results corresponding to the text recognition result can be to mark the word vector recognition results of the word vector corresponding to the effective text in the text recognition result as effective text, and to mark the word vector recognition results of the word vector corresponding to the non-effective text in the text recognition result as non-effective text.

[0106] Step 930, establish an initial network model, and train the initial network model based on multiple word vector samples and the word vector recognition results corresponding to the multiple word vector samples to generate an efficacy recognition model.

[0107] The efficacy recognition model mentioned in step 930 is used to generate a word vector recognition result corresponding to the input word vector based on the input word vector. The initial network model can be a BERT model. BERT is an open source pre-trained language model. The pre-trained language model is trained on a wide range of data sets. Therefore, the BERT model is a model that already has certain basic language prior knowledge. The BERT model is used as the initial network model for training. During the training process, the BERT model continuously adjusts the parameters in the model framework. Through continuous iterative adjustment, the BERT model achieves the optimal effect, improves the learning efficiency of the initial network model, and at the same time improves the accuracy of the efficacy recognition model.

[0108] Fig.10 FIG. 1 is a flowchart of a text recognition method provided by another embodiment of the present application. Fig. 9 The present application is extended based on the embodiment shown Fig.10 The embodiment shown is described below in detail. Fig.10 The differences between the illustrated embodiments and Fig. 9 the illustrated embodiments will not be elaborated here as the similarities are already described. As Fig.10 shown, the steps of determining the training text and the text recognition result corresponding to the training text include the following steps.

[0109] Step 1010: Obtain the first-language training text and the second-language training text.

[0110] Step 1020: Based on the first-language training text and the first text recognition result, obtain the first efficacy-labeled text corresponding to the first-language training text.

[0111] Exemplarily, the first efficacy-labeled text may be the efficacy-labeled text.

[0112] Exemplarily, the training text includes the first-language training text and the second-language training text. The first-language training text and the second-language training text include equivalent content written in different languages respectively. The first-language training text includes the text content written in the first language, and the second-language training text includes the text content written in the second language. The text recognition result corresponding to the first-language training text is the first text recognition result, and the text recognition result corresponding to the second-language training text is the second text recognition result. The text recognition result includes the first text recognition result and the second text recognition result.

[0113] In an embodiment of the present application, the training text is a patent text, and the first-language training text and the second-language training text belong to the same patent family. In an embodiment of the present application, the first language includes Japanese, and the first-language training text includes Japanese patent texts.

[0114] Specifically, most Japanese patent texts are marked with the label of "invention effect", for example, marked with "発明の効果". In practical applications, Japanese patent texts can be screened out first, and then Japanese patent texts marked with "invention effect" can be screened out. The first efficacy-labeled text may be the text of the effect part marked with "invention effect".

[0115] Step 1030: Based on the first efficacy-labeled text, determine the second text recognition result corresponding to the second-language training text.

[0116] By determining the first-language training text and the second-language training text, and based on the first efficacy-labeled text of the first-language training text, determining the second text recognition result of the second-language training text, it is possible to quickly determine the second text recognition result for the second-language training text that originally did not have the second text recognition result, improving the determination efficiency of the training text and the text recognition result corresponding to the training text.

[0117] Fig.11aThe following is a schematic flowchart of a text recognition method provided by another embodiment of the present application. Based on the embodiment of the present application Fig.10 the embodiment shown is extended to form the embodiment of the present application Fig.11a shown below. The differences between the embodiment shown Fig.11a and the embodiment shown Fig.10 will be mainly described below. The same parts will not be elaborated. As Fig.11a shown, the step of determining the second text recognition result corresponding to the second language training text based on the first efficacy-marked text includes the following steps.

[0118] Step 1110: Translate the first efficacy-marked text to obtain the translated text corresponding to the first efficacy-marked text.

[0119] Exemplarily, the translated text is expressed in the second language. The first language can be Japanese. As Fig.11b shown, the first efficacy-marked text in the first language can be the text of the effect part marked with "発明の効果". The second language can be Chinese, English, or other languages, which is not specifically limited in the present application.

[0120] Step 1120: Use a similarity algorithm to determine the second text recognition result based on the translated text corresponding to the first efficacy-marked text.

[0121] By calculating the similarity between the first efficacy-marked text in the second language and the second language training text, the sentence or paragraph in the second language training text with the highest similarity to the first efficacy-marked text in the second language can be obtained, that is, the sentence or paragraph with the highest similarity to the first efficacy-marked text in the second language is the second text recognition result of the second language training text, thereby quickly determining the second text recognition result and improving the determination efficiency of the training text and the text recognition result corresponding to the training text.

[0122] In an embodiment of the present application, the step of translating the first efficacy-marked text to obtain the translated text corresponding to the first efficacy-marked text includes the following steps.

[0123] Input the first efficacy-marked text into a text translation model to generate the translated text corresponding to the first efficacy-marked text.

[0124] By using a text translation model to translate the first efficacy-marked text in the first language into the first efficacy-marked text in the second language, the efficiency of text recognition is further improved.

[0125] Fig.12 The following is a schematic flowchart of a text recognition method provided by another embodiment of the present application. Based on the embodiment of the present application Fig.10 the embodiment shown is extended to form the embodiment of the present application Fig.12 shown below. The following will mainly describe Fig.12 The embodiment shown and Fig.10 The differences between the embodiments shown are shown, and the similarities are not repeated. Fig.12 As shown, the step of acquiring the first language training text and the second language training text includes the following steps.

[0126] Step 1210, obtaining patent family text data in multiple different languages.

[0127] Exemplarily, the patent family text data may be simple family text data.

[0128] Step 1220 , screening the first language training text and the second language training text based on the patent family text data.

[0129] Exemplarily, the first language training text includes efficacy identification paragraph information, and the second language training text includes patent text to be efficacy marked.

[0130] In practical applications, taking the first language training text as a Japanese patent and the second language training text as a Chinese patent as an example, the training text can be obtained by the following steps:

[0131] (1) Multiple groups of family patents can be screened out from the patent database, each group of family patents including Japanese patents and Chinese patents.

[0132] (2) Filter out the Japanese patents in all family patents, and then filter out the Japanese patent texts marked with "invention effect" from all Japanese patents.

[0133] (3) Translate the part marked with “Effects of the Invention” into Chinese text.

[0134] (4) Calculate the similarity between the translated Chinese text and the entire Chinese patent, and obtain the part of the entire Chinese patent that has a higher similarity with the translated Chinese text as the text recognition result.

[0135] Fig.13 FIG. 1 is a flowchart of a text recognition method provided by another embodiment of the present application. Fig.10 The present application is extended based on the embodiment shown Fig.13 The embodiment shown is described below in detail. Fig.13 The embodiment shown and Fig.10 The differences between the embodiments shown are shown, and the similarities are not repeated. Fig.13 As shown, after the step of determining the second text recognition result corresponding to the second language training text based on the first efficacy marked text, the following steps are also included.

[0136] Step 1310: Determine a second efficacy mark text corresponding to the second language training text.

[0137] Specifically, the second language training text may also include a second effect marking text. For example, some Chinese patents may also include text with the effect part marked with "inventive effect". The second effect marking text may be text with the effect part marked with "inventive effect", i.e., effect marking text.

[0138] Step 1320: Determine the accuracy of the second text recognition result based on the second text recognition result and the second efficacy mark text.

[0139] Specifically, accuracy can be precision A or recall R.

[0140] The accuracy is calculated as follows:

[0141]

[0142] The calculation formula for recall is as follows:

[0143]

[0144] Among them, W1 is the number of functional paragraphs or functional sentences in the second text recognition results of the second language training text in all training texts, and W2 is the number of functional paragraphs or functional sentences in the second functional marking text of the second language training text in all training texts.

[0145] The present application has obtained through comparative experiments that the precision and recall rate obtained by the method of the present application are both greater than 90%. In addition, the method of the present application only requires hundreds of thousands of training texts to achieve satisfactory results.

[0146] Step 1330: Adjust the training text according to the accuracy of the second text recognition result.

[0147] Specifically, if it is found that the accuracy is not high when determining the training text, for example, the precision and the recall rate are both less than 80%, the training text may be adjusted, for example, the training text may be reselected.

[0148] The training text is adjusted by calculating the accuracy of the second text recognition result, thereby improving the quality of the training text, thereby improving the quality of the efficacy recognition model and further improving the accuracy of text recognition.

[0149] Exemplary Devices

[0150] Combination of the above Figures 2 to 13 , describes the method embodiment of the present application in detail, and the following is combined with Figures 14 to 24, describes the device embodiment of the present application in detail. The description of the method embodiment corresponds to the description of the device embodiment, so the parts not described in detail can refer to the previous method embodiment.

[0151] Fig.14 Shown is a structural schematic diagram of a text recognition device provided in one embodiment of the present application.

[0152] like Fig.14 As shown, the text recognition device 1400 provided in the embodiment of the present application includes:

[0153] The splitting module 1410 is configured to determine, based on the text to be recognized, a plurality of character strings corresponding to the text to be recognized, wherein adjacent character strings in the plurality of character strings partially overlap;

[0154] The conversion module 1420 is configured to perform word vector conversion on the multiple character strings to obtain multiple word vectors, wherein the multiple word vectors are in a one-to-one correspondence with the multiple character strings;

[0155] A generating module 1430 is configured to generate a plurality of word vector recognition results corresponding to each of the word vectors based on the plurality of word vectors, wherein the word vector recognition results are functional texts or non-functional texts;

[0156] The determination module 1440 is configured to determine the text recognition result of the text to be recognized based on the word vector recognition results corresponding to each of the multiple word vectors.

[0157] Fig.15 The figure shows a schematic diagram of the structure of a character string determination unit provided in one embodiment of the present application. Fig.14 The present application is extended based on the embodiment shown Fig.15 The embodiment shown is described below in detail. Fig.15 The embodiment shown and Fig.14 The differences and similarities of the illustrated embodiments are not described in detail.

[0158] like Fig.15 As shown, in the text recognition device 1400 provided in the embodiment of the present application, the splitting module 1410 includes:

[0159] A first character string determining unit 1511 is configured to determine a first character string of a preset character string length based on the text to be recognized;

[0160] The Nth character string determining unit 1512 is configured to use the characters in the Nth character string as the starting characters of the N+1th character string, decompose the text to be recognized based on a preset character string length and a preset decomposition step length, and obtain the N+1th character string, where N is a positive integer greater than or equal to 1 and less than M.

[0161] Fig.16The figure shows a schematic diagram of the structure of a determination module provided in one embodiment of the present application. Fig.14 The present application is extended based on the embodiment shown Fig.16 The embodiment shown is described below in detail. Fig.16 The embodiment shown and Fig.14 The differences and similarities of the illustrated embodiments are not described in detail.

[0162] like Fig.16 As shown, in the text recognition device 1400 provided in the embodiment of the present application, the determination module 1440 includes:

[0163] The voting determination unit 1441 is configured to adopt a voting mechanism to perform a voting operation on the word vector recognition results corresponding to each of the multiple word vectors to determine the text recognition result of the text to be recognized.

[0164] Fig.17 The figure shows a schematic diagram of the structure of a voting unit provided in one embodiment of the present application. Fig.16 The present application is extended based on the embodiment shown Fig.17 The embodiment shown is described below in detail. Fig.17 The embodiment shown and Fig.16 The differences and similarities of the illustrated embodiments are not described in detail.

[0165] like Fig.17 As shown, in the text recognition device 1400 provided in the embodiment of the present application, the voting determination unit 1441 includes:

[0166] The pre-voting processing subunit 1711 is configured to determine a plurality of to-be-recognized units corresponding to the to-be-recognized text based on the to-be-recognized text, wherein the to-be-recognized unit corresponds to at least one word vector;

[0167] The voting subunit 1712 is configured to use a voting mechanism to vote on the word vector recognition result corresponding to each unit to be recognized among the multiple units to be recognized, so as to determine the text recognition result of the unit to be recognized;

[0168] The post-voting processing subunit 1713 is configured to determine the text recognition result corresponding to the text to be recognized based on the text recognition results corresponding to each of the multiple units to be recognized.

[0169] Fig.18 The figure is a schematic diagram of the structure of a voting operation subunit provided in one embodiment of the present application. Fig.17 The present application is extended based on the embodiment shown Fig.18 The embodiment shown is described below in detail. Fig.18 The embodiment shown and Fig.17 The differences and similarities of the illustrated embodiments are not described in detail.

[0170] like Fig.18 As shown, in the text recognition device 1400 provided in the embodiment of the present application, the voting subunit 1712 includes:

[0171] The voting operation subunit 1811 is configured to determine that for each unit to be recognized among the multiple units to be recognized, if the number of effective texts in the word vector recognition result corresponding to the unit to be recognized is greater than or equal to the number of non-effective texts, the text recognition result of the unit to be recognized is determined to be effective text.

[0172] Fig.19 The figure shows a schematic diagram of the structure of a generation module provided in one embodiment of the present application. Fig.14 The present application is extended based on the embodiment shown Fig.19 The embodiment shown is described below in detail. Fig.19 The embodiment shown and Fig.14 The differences and similarities of the illustrated embodiments are not described in detail.

[0173] like Fig.19 As shown, in the text recognition device 1400 provided in the embodiment of the present application, the generating module 1430 includes:

[0174] The generating unit 1431 is configured to generate a plurality of word vector recognition results corresponding to each word vector based on a plurality of word vectors using an efficacy recognition model, wherein the efficacy recognition model is used to generate a word vector recognition result corresponding to an input word vector based on an input word vector.

[0175] Fig. 20 The figure shows a structural diagram of a text recognition device provided by another embodiment of the present application. Fig.19 The present application is extended based on the embodiment shown Fig. 20 The embodiment shown is described below in detail. Fig. 20 The embodiment shown and Fig.19 The differences and similarities of the illustrated embodiments are not described in detail.

[0176] like Fig. 20 As shown, the text recognition device 1400 provided in the embodiment of the present application further includes:

[0177] A sample preprocessing module 1450 is configured to determine a training text and a text recognition result corresponding to the training text;

[0178] The sample determination module 1460 is configured to determine, based on the training text and the text recognition result corresponding to the training text, a plurality of word vector samples corresponding to the training text and the word vector recognition results corresponding to each of the plurality of word vector samples;

[0179] The model determination module 1470 is configured to establish an initial network model and train the initial network model based on multiple word vector samples and word vector recognition results corresponding to each of the multiple word vector samples to generate an efficacy recognition model.

[0180] Fig.21 The figure shows a schematic diagram of the structure of a sample preprocessing module provided in one embodiment of the present application. Fig. 20 The present application is extended based on the embodiment shown Fig.21 The embodiment shown is described below in detail. Fig.21 The embodiment shown and Fig. 20 The differences and similarities of the illustrated embodiments are not described in detail.

[0181] like Fig.21 As shown, in the text recognition device 1400 provided in the embodiment of the present application, the sample preprocessing module 1450 includes:

[0182] A first sample preprocessing unit 1451 is configured to obtain a first language training text and a second language training text;

[0183] The second sample preprocessing unit 1452 obtains a first efficacy mark text corresponding to the first language training text based on the first language training text and the first text recognition result;

[0184] The third sample preprocessing unit 1453 is configured to determine a second text recognition result corresponding to the second language training text based on the first efficacy mark text.

[0185] Fig. 22 The figure shows a schematic diagram of the structure of a second sample preprocessing unit provided in one embodiment of the present application. Fig.21 The present application is extended based on the embodiment shown Fig. 22 The embodiment shown is described below in detail. Fig. 22 The embodiment shown and Fig.21 The differences and similarities of the illustrated embodiments are not described in detail.

[0186] like Fig. 22 As shown, in the text recognition device 1400 provided in the embodiment of the present application, the second sample preprocessing unit 1452 includes:

[0187] The translation subunit 2210 is configured to translate the first effect mark text to obtain a translation text corresponding to the first effect mark text, wherein the translation text is expressed in a second language;

[0188] The similarity determination subunit 2220 is configured to use a similarity algorithm to determine a second text recognition result based on the translation text corresponding to the first efficacy mark text.

[0189] The translation subunit 2210 is further configured to input the first efficacy mark text into a text translation model to generate a translation text corresponding to the first efficacy mark text.

[0190] Fig.23 The figure shows a schematic diagram of the structure of a sample preprocessing module provided by another embodiment of the present application. Fig.21 The present application is extended based on the embodiment shown Fig.23 The embodiment shown is described below in detail. Fig.23 The embodiment shown and Fig.21 The differences and similarities of the illustrated embodiments are not described in detail.

[0191] like Fig.23 As shown, in the text recognition device 1400 provided in the embodiment of the present application, the sample preprocessing module 1450 further includes:

[0192] The fourth sample preprocessing unit 1454 is configured to determine a second efficacy mark text corresponding to the second language training text;

[0193] A comparison unit 1455, configured to determine the accuracy of the second text recognition result based on the second text recognition result and the second efficacy mark text;

[0194] The adjusting unit 1456 is configured to adjust the training text according to the accuracy of the second text recognition result.

[0195] Fig.24 The figure shows a schematic diagram of the structure of a first sample preprocessing unit provided in another embodiment of the present application. Fig.21 The present application is extended based on the embodiment shown Fig.24 The embodiment shown is described below in detail. Fig.24 The embodiment shown and Fig.21 The differences and similarities of the illustrated embodiments are not described in detail.

[0196] like Fig.24 As shown, in the text recognition device 1400 provided in the embodiment of the present application, the first sample preprocessing unit 1451 further includes:

[0197] The sample acquisition subunit 2410 is configured to acquire patent family text data in multiple different languages.

[0198] The screening subunit 2420 is configured to screen the first language training text and the second language training text based on the patent family text data.

[0199] Exemplary Electronic Devices

[0200] Fig.25 FIG. 1 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Fig.25 As shown, the electronic device 250 includes: one or more processors 2501 and a memory 2502; and computer program instructions stored in the memory 2502, and when the computer program instructions are executed by the processor 2501, the processor 2501 executes the text recognition method as any of the above embodiments.

[0201] The processor 2501 may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.

[0202] The memory 2502 may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory (cache), etc. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage medium, and the processor 2501 may execute the program instructions to implement the steps in the text recognition method of each embodiment of the present application described above and / or other desired functions.

[0203] In one example, the electronic device 250 may further include: an input device 2503 and an output device 2504, which are connected via a bus system and / or other forms of connection mechanisms ( Fig.25 ) interconnected.

[0204] In addition, the input device 2503 may also include, for example, a keyboard, a mouse, a microphone, etc.

[0205] The output device 2504 can output various information to the outside, and may include, for example, a display, a speaker, a printer, a communication network and its connected remote output devices, etc.

[0206] Of course, to simplify, Fig.25 Only some of the components related to the present application in the electronic device 250 are shown, and components such as a bus, an input device / output interface, etc. are omitted. In addition, according to specific application conditions, the electronic device 250 may also include any other appropriate components.

[0207] Exemplary computer-readable storage media

[0208] In addition to the above methods and devices, the embodiments of the present application may also be a computer program product, including computer program instructions, which, when executed by a processor, enable the processor to execute the steps in the text recognition method of any of the above embodiments.

[0209] The computer program product may be written in any combination of one or more programming languages ​​to write program codes for performing the operations of the embodiments of the present application, including object-oriented programming languages ​​such as Java, C++, etc., and conventional procedural programming languages ​​such as "C" language or similar programming languages. The program code may be executed entirely on the user computing device, partially on the user device, as a separate software package, partially on the user computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0210] In addition, an embodiment of the present application may also be a computer-readable storage medium on which computer program instructions are stored. When the computer program instructions are executed by a processor, the processor executes the steps of the text recognition method according to various embodiments of the present application described in the above "Exemplary Method" section of this specification.

[0211] Computer readable storage media can adopt any combination of one or more readable media. The readable medium can be a readable signal medium or a readable storage medium. The readable storage medium can include, for example, but is not limited to, a system, device or device of electricity, magnetism, light, electromagnetic, infrared, or semiconductor, or any combination of the above. More specific examples (non-exhaustive list) of readable storage media include: an electrical connection with one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0212] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.

[0213] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0214] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0215] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0216] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

[0217] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A text recognition method, characterized in that: include: Determine, based on the text to be recognized, a plurality of character strings corresponding to the text to be recognized, wherein adjacent character strings in the plurality of character strings partially overlap, and the text to be recognized is a patent text; Performing word vector conversion on the multiple character strings to obtain multiple word vectors, wherein the multiple word vectors are in a one-to-one correspondence with the multiple character strings; Generating word vector recognition results corresponding to each of the multiple word vectors based on the multiple word vectors, wherein the word vector recognition results are efficacy texts or non-efficacy texts, and the efficacy texts are texts representing technical effects; and Determine a text recognition result of the text to be recognized based on the word vector recognition results corresponding to each of the multiple word vectors; The step of generating the word vector recognition results corresponding to each of the multiple word vectors based on the multiple word vectors includes: Determine a training text and a text recognition result corresponding to the training text, the training text includes a first language training text and a second language training text, the first language training text includes text content written in a first language, the second language training text includes the text content written in a second language, the text recognition result corresponding to the first language training text is a first text recognition result, and the text recognition result corresponding to the second language training text is a second text recognition result; Based on the training text and the text recognition result corresponding to the training text, determine a plurality of word vector samples corresponding to the training text and the word vector recognition results corresponding to each of the plurality of word vector samples; Establishing an initial network model, and training the initial network model based on the multiple word vector samples and the word vector recognition results corresponding to the multiple word vector samples to generate an efficacy recognition model, wherein the efficacy recognition model is used to generate a word vector recognition result corresponding to the input word vector based on the input word vector; Using the efficacy recognition model, generating word vector recognition results corresponding to each of the multiple word vectors based on the multiple word vectors; The step of determining the training text and the text recognition result corresponding to the training text includes: Acquire the first language training text and the second language training text, wherein the first language training text and the second language training text belong to the same patent family, the first language training text includes efficacy identification paragraph information, and the second language training text includes patent text to be efficacy marked; Based on the first language training text and the first text recognition result, obtaining a first efficacy mark text corresponding to the first language training text; translating the first efficacy mark text to obtain a translation text corresponding to the first efficacy mark text, wherein the translation text is expressed in the second language; Using a similarity algorithm, based on the translation text corresponding to the first efficacy mark text, a sentence or paragraph in the second language training text having the highest similarity to the translation text corresponding to the first efficacy mark text is determined as the second text recognition result.

2. The text recognition method according to claim 1, characterized in that: The number of the multiple character strings is M, where M is a positive integer greater than 1, and the determining, based on the text to be recognized, the multiple character strings corresponding to the text to be recognized includes: Determine a first character string of a preset character string length based on the text to be recognized; Taking the characters in the Nth character string as the starting characters of the N+1th character string, decomposing the text to be recognized based on the preset character string length and the preset decomposition step length, to obtain the N+1th character string, where N is a positive integer greater than or equal to 1 and less than M.

3. The text recognition method according to claim 1, characterized in that: The determining the text recognition result of the to-be-recognized text based on the word vector recognition results corresponding to each of the multiple word vectors includes: A voting mechanism is adopted to perform a voting operation on the word vector recognition results corresponding to each of the multiple word vectors to determine the text recognition result of the text to be recognized.

4. The text recognition method according to claim 3, characterized in that: The adopting of a voting mechanism to perform a voting operation on the word vector recognition results corresponding to each of the multiple word vectors to determine the text recognition result of the text to be recognized includes: Determine, based on the text to be recognized, a plurality of units to be recognized corresponding to the text to be recognized, wherein the unit to be recognized corresponds to at least one of the word vectors; For each unit to be recognized among the multiple units to be recognized, the voting mechanism is used to perform the voting operation on the word vector recognition result corresponding to the unit to be recognized, so as to determine the text recognition result of the unit to be recognized; Based on the text recognition results corresponding to each of the multiple units to be recognized, a text recognition result corresponding to the text to be recognized is determined.

5. The text recognition method according to claim 4, characterized in that: The unit to be identified includes at least one of a sentence, a paragraph and a text module in the patent text; wherein the text module includes at least one of an abstract module, a claim module and a specification module.

6. The text recognition method according to claim 4, characterized in that: The adopting the voting mechanism to perform the voting operation on the word vector recognition result corresponding to the unit to be recognized to determine the text recognition result of the unit to be recognized includes: If, in the word vector recognition result corresponding to the unit to be recognized, the number of the functional texts is greater than or equal to the number of the non-functional texts, it is determined that the text recognition result of the unit to be recognized is the functional text.

7. The text recognition method according to claim 1, characterized in that: The training text is a patent text, and obtaining the first language training text and the second language training text includes: Access patent family text data in multiple languages; The first language training text and the second language training text are screened based on the patent family text data.

8. The text recognition method according to claim 7, characterized in that: The first language includes Japanese, and the first language training text includes Japanese patent text.

9. A text recognition device, characterized in that: include: A splitting module, configured to determine, based on the text to be recognized, a plurality of character strings corresponding to the text to be recognized, wherein adjacent character strings in the plurality of character strings partially overlap, and the text to be recognized is a patent text; A conversion module, configured to perform word vector conversion on the multiple character strings to obtain multiple word vectors, wherein the multiple word vectors are in a one-to-one correspondence with the multiple character strings; A generating module configured to generate word vector recognition results corresponding to each of the plurality of word vectors based on the plurality of word vectors, wherein the word vector recognition results are efficacy texts or non-efficacy texts, and the efficacy texts are texts representing technical effects; and A determination module, configured to determine a text recognition result of the to-be-recognized text based on the word vector recognition results corresponding to each of the plurality of word vectors; The step of generating the word vector recognition results corresponding to each of the multiple word vectors based on the multiple word vectors includes: Determine a training text and a text recognition result corresponding to the training text, the training text includes a first language training text and a second language training text, the first language training text includes text content written in a first language, the second language training text includes the text content written in a second language, the text recognition result corresponding to the first language training text is a first text recognition result, and the text recognition result corresponding to the second language training text is a second text recognition result; Based on the training text and the text recognition result corresponding to the training text, determine a plurality of word vector samples corresponding to the training text and the word vector recognition results corresponding to each of the plurality of word vector samples; Establishing an initial network model, and training the initial network model based on the multiple word vector samples and the word vector recognition results corresponding to the multiple word vector samples to generate an efficacy recognition model, wherein the efficacy recognition model is used to generate a word vector recognition result corresponding to the input word vector based on the input word vector; Using the efficacy recognition model, generating word vector recognition results corresponding to each of the multiple word vectors based on the multiple word vectors; The step of determining the training text and the text recognition result corresponding to the training text includes: Acquire the first language training text and the second language training text, wherein the first language training text and the second language training text belong to the same patent family, the first language training text includes efficacy identification paragraph information, and the second language training text includes patent text to be efficacy marked; Based on the first language training text and the first text recognition result, obtaining a first efficacy mark text corresponding to the first language training text; translating the first efficacy mark text to obtain a translation text corresponding to the first efficacy mark text, wherein the translation text is expressed in the second language; Using a similarity algorithm, based on the translation text corresponding to the first efficacy mark text, a sentence or paragraph in the second language training text having the highest similarity to the translation text corresponding to the first efficacy mark text is determined as the second text recognition result.

10. A computer-readable storage medium, characterized in that: The storage medium stores instructions, and when the instructions are executed by a processor of an electronic device, the electronic device is enabled to execute the text recognition method described in any one of claims 1 to 8.

11. An electronic device, characterized in that: The electronic device comprises: processor; memory for storing computer executable instructions; The processor is used to execute the computer executable instructions to implement the text recognition method described in any one of claims 1 to 8.

Images