A method for obtaining a prediction model, an electronic device and a storage medium

Abstract

The present invention provides a method for obtaining a prediction model, an electronic device and a storage medium. The method comprises the following steps: obtaining a plurality of target texts from a preset text library according to a plurality of target word objects under a category to which a given word object belongs, then obtaining a score influence indicator set corresponding to each target text, classifying the plurality of target texts according to text type to obtain a plurality of target text sets, performing unsupervised clustering on each target text set to obtain a plurality of key texts, obtaining a technology maturity index according to the score influence indicator set corresponding to each key text and a preset life cycle indicator model, and obtaining a technology maturity prediction model corresponding to the given word object according to the technology maturity index and a pre-acquired target growth limit; the technology maturity prediction model obtained by the present invention has a more accurate and reliable prediction accuracy, and is helpful to understand the development level of the technology.

Description

Technical Field

[0001] The present invention relates to the field of text data, and in particular to a method for acquiring a prediction model, an electronic device and a storage medium. Background Art

[0002] Technological innovation is defined as a technology system that requires long-term high-investment research and development and has key and unique characteristics. It is of key significance for breaking through the technical bottlenecks of the entire industrial development. Key core technologies themselves have a high degree of external dependence. Basic processes, key materials, core equipment, and technical routes are highly dependent on the supply of other companies and the support of other industrial links. In the process of development, enterprises or industries have high technological dependence or external dependence. Therefore, it is very meaningful to master the development level and development trend of technology for future decision-making. However, for many technologies, the development level is only predicted by relevant technicians based on their work experience, which has low reliability and is not a good reference. Summary of the invention

[0003] In view of the above technical problems, the technical solution adopted by the present invention is:

[0004] A method for obtaining a prediction model, characterized in that the method comprises the following steps:

[0005] According to a plurality of target word objects in the category to which the given word object belongs that are acquired in advance, a plurality of target texts corresponding to the given word object are acquired from a preset text library; the given word object is the name of any technology specified by the user.

[0006] Obtaining a score impact indicator set corresponding to each target text; the score impact indicator set corresponding to the target text includes a normalized value of a duration parameter corresponding to the target text, a normalized value of the number of target word objects, and an average priority of the target word objects;

[0007] A number of target texts are classified according to text types to obtain a number of target text sets.

[0008] Unsupervised clustering is performed on each target text set to obtain a number of key texts, thereby obtaining a score influence indicator set corresponding to each key text; the key text refers to the target text corresponding to the center of any target text cluster obtained by unsupervised clustering.

[0009] The normalized value of the duration parameter in the score impact index set corresponding to the i-th key text, the normalized value of the number of target word objects, the average priority of the target word objects, and the preset score corresponding to the i-th key text are taken as x i1 、x i2 、x i3 and Z iSubstitute into the preset life cycle indicator model Z i =ax i1 +bx i2 +cx i3 +d, the to-be-processed lifecycle indicator model corresponding to the i-th key text is obtained, where a, b, c, and d are constants.

[0010] According to the pending life cycle indicator model corresponding to each key text, the values ​​of a, b, c, and d are obtained through a genetic algorithm to obtain the target life cycle indicator model, and the target life cycle indicator model is used as the technology maturity index.

[0011] According to the technology maturity index and the pre-acquired target growth limit, a technology maturity prediction model corresponding to a given word object is obtained.

[0012] The present invention has at least the following beneficial effects:

[0013] The present invention provides a method for obtaining a prediction model. First, according to a plurality of target word objects under a category to which a given word object belongs, a plurality of target texts including the target word objects are obtained from a preset text library, and then a score influence indicator set corresponding to each target text is obtained. The plurality of target texts are classified according to text types to obtain a plurality of target text sets, and unsupervised clustering is performed on each target text set to obtain a plurality of key texts. Through unsupervised clustering, the center of each target text cluster is screened out as the screened key text, so that the number of used texts can be reduced and the subsequent calculation complexity is greatly reduced. Then, according to the score influence indicator set corresponding to each key text and a preset life cycle indicator model, a technology maturity index is obtained. According to the technology maturity index and a pre-acquired target growth limit, a technology maturity prediction model corresponding to the given word object is obtained. Since the score influence indicator is closely related to the given technology, the three indicators are respectively used as variables in the preset life cycle indicator model and a series of calculations are performed, so that the obtained technology maturity prediction model has a more accurate and reliable prediction accuracy, which is helpful to understand the development level of the technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0015] Figure 1 A flowchart of a method for obtaining a prediction model provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0017] This embodiment provides a method for obtaining a prediction model, such as Figure 1 As shown, the method comprises the following steps:

[0018] S100, acquiring a plurality of target texts corresponding to the given word object from a preset text library according to a plurality of target word objects in the category to which the given word object belongs that are acquired in advance; the given word object is the name of any technology specified by the user.

[0019] In a specific embodiment, the target text corresponding to a given word object is obtained through the following steps:

[0020] S101, based on a plurality of target word objects in a category to which a given word object belongs that are obtained in advance, a plurality of first texts to be processed corresponding to each target word object are obtained from a preset text library, and a plurality of target sentences are determined from each first text to be processed; the target sentence refers to a sentence including at least one target word object obtained from a target paragraph in any first text to be processed.

[0021] In a specific embodiment, before step S101, several target word objects under the category to which the given word object belongs are obtained by the following steps:

[0022] S001, adding a given word object to a target position in a preset query text template to generate a target query text; the target position refers to a position in the query text template where a missing text exists.

[0023] For example, the query text template is "What are the words related to —— technology?", if the given word object is image processing, add image processing to "——".

[0024] S002, input the target query text into a preset large language model to obtain a number of target word objects under the category to which the given word object belongs. For example, the target word object can be a synonym, an alias, a subordinate word of the same genus, etc. of the given word object. The preset large language model is a model trained in advance for obtaining a number of word objects related to any word object. Those skilled in the art know the specific training process, which will not be described in detail here.

[0025] In a specific embodiment, step S101 further includes the following steps:

[0026] S1011, obtaining from a preset text library a plurality of first texts to be processed corresponding to each target word object and a text category identifier of each first text to be processed; this can be understood as: obtaining from the preset text library a text containing at least one target word object as the first text to be processed.

[0027] Specifically, the text category identification includes but is not limited to patent text and paper text.

[0028] S1012, for any first text to be processed, based on the correspondence between the preset text category identifier and the target paragraph in the first text to be processed, according to the text category identifier of the first text to be processed, determine the target paragraph from the first text to be processed; it can be understood that: the correspondence between the preset text category identifier and the target paragraph in the first text to be processed includes a plurality of groups of one-to-one corresponding text category identifiers and target paragraphs. For example, when the text category identifier is a patent text, considering the repetitiveness of words, the specific implementation method part is used as the target paragraph.

[0029] As described above, by obtaining the first text to be processed based on a number of target word objects, the obtained related text is made more comprehensive, and then the target sentence including the target word object is further determined from the target paragraph of the first text to be processed, so that the number of each target word object in the obtained target sentence can provide a basis for calculating the text priority of the first text to be processed.

[0030] S102: For any first text to be processed, according to the quantity information of each target word object in each target sentence in the first text to be processed, obtain the first text priority corresponding to the first text to be processed.

[0031] In a specific embodiment, the first text priority corresponding to the first to-be-processed text is obtained through the following steps:

[0032] S1021, for any target sentence in the first text to be processed, divide a number of target word objects in the target sentence into a number of target phrases; the target phrase refers to any phrase in which the number of target word objects screened out from the currently remaining number of target word objects is the largest and each screened out target word object has only one; the currently remaining number of target word objects refers to the remaining number of target word objects obtained after deleting all the screened out target word objects from the number of target word objects in the target sentence.

[0033] For a better understanding, an example is given as follows: when the target word objects in the target sentence include A1, A2, A3, A2, A1, A1, A1, the first screening obtains the target phrase (A1, A2, A3), and the remaining target word objects currently include A2, A1, A1, A1, and the second screening obtains the target phrase (A1, A2). The remaining target word objects currently include A1, A1, and the third screening obtains the target phrase (A1). The remaining target word objects currently are A1, so the last screening obtains the target phrase (A1).

[0034] S1022, for any target phrase corresponding to the target sentence, calculate the first phrase priority corresponding to the target phrase based on the total number of target word objects in the target phrase and the target number corresponding to the target phrase; the target number corresponding to the target phrase refers to the number of phrases obtained from several target phrases corresponding to the target sentence that are the same as any target phrase.

[0035] Further, the first phrase priority corresponding to the target phrase is the product of the total number of target word objects in the target phrase and the target number corresponding to the target phrase.

[0036] For example, when the target phrase is (A1, A2, A3), the total number of target word objects in the target phrase is 3, the target number corresponding to the target phrase is 1, and the first phrase priority corresponding to the target phrase is 3; when the target phrase is (A1), the total number of target word objects in the target phrase is 1, the target number corresponding to the target phrase is 2, and the first phrase priority corresponding to the target phrase is 2.

[0037] S1023, based on the same target phrase information in the several target phrases corresponding to the target sentence, deduplicate the several first phrase priorities to obtain several second phrase priorities; it can be understood as: deduplicate the target phrases, and also remove the first phrase priorities corresponding to the removed target phrases, because when calculating the first phrase priority of the target phrase, all target phrases identical to the target phrase have been taken into account.

[0038] S1024: Determine the sum of the priorities of the plurality of second phrases as the sentence priority corresponding to the target sentence.

[0039] S1025, adding sentence priorities corresponding to a plurality of target sentences in the first text to be processed, to obtain a first text priority corresponding to the first text to be processed.

[0040] As described above, when calculating the first text priority of the first text to be processed, the dimension of the target word objects contained in the first text to be processed is considered. By introducing the combination of target word objects and the number of each combination, the importance of the first text to be processed can be more accurately reflected.

[0041] S103, obtaining a plurality of second texts to be processed corresponding to the first text to be processed, and obtaining the second text priority corresponding to the first text to be processed according to a plurality of target word objects corresponding to each second text to be processed and the duration between the publication time of the first text to be processed and the publication time of each second text to be processed.

[0042] Specifically, the second text to be processed refers to any other first text to be processed that is published earlier than the first text to be processed. For example, the target word object corresponds to five first texts to be processed, which are arranged in order of publication time. When the first text to be processed is the fourth published text, the corresponding second texts to be processed are the three first texts to be processed that were published before the first text to be processed itself.

[0043] In a specific embodiment, the second text priority corresponding to the first to-be-processed text is obtained by the following steps:

[0044] S1031, according to the time between the publication time of the first to-be-processed text and the publication time of each second to-be-processed text, normalize each time to obtain a time proportion parameter corresponding to each second to-be-processed text.

[0045] S1032, for any second text to be processed, according to the target word objects corresponding to the second text to be processed, obtain the number of repeated target word objects between the second text to be processed and the first text to be processed. For example, the first text to be processed includes target word objects A1, A2, A3, and the second text to be processed includes target word objects A1 and A2, then the total number of repeated target word objects is 2.

[0046] S1033: Determine the product of the time proportion parameter corresponding to the second text to be processed and the total number of target word objects repeated between the second text to be processed and the first text to be processed as the text weight of the second text to be processed.

[0047] S1034, calculating the ratio of the sum of the text weights of each second text to be processed to the number of the second texts to be processed, and using the calculation result as the second text priority corresponding to the first text to be processed.

[0048] As described above, when calculating the second text priority of the first text to be processed, the dimension of several second texts to be processed whose publication time is before the first text to be processed is taken into consideration, and the time dimension is also introduced into the number of repeated words as the weight of the number of repeated words, which can accurately reflect the importance of each second text to be processed to the first text to be processed, thereby improving the reliability of the second text priority corresponding to the first text to be processed.

[0049] S104, when the sum of the first text priority and the corresponding second text priority corresponding to the first text to be processed is greater than the preset priority threshold, the first text to be processed is used as the target text corresponding to the given word object; it can be understood that: the target text corresponding to the given word object refers to the relevant research text of the technology to which the given word object belongs.

[0050] As described above, the priority of the text is considered through multiple dimensions, which makes the determination of the target text with high relevance to the technology to which the given word object belongs more reasonable and reliable, and the several target texts with different disclosure times that are finally determined are conducive to reflecting the development process of the technology to a certain extent, and provide a strong basis for further predicting the life cycle of the technology to which the given word object belongs.

[0051] S200, obtaining a score influence indicator set corresponding to each target text; the score influence indicator set corresponding to the target text includes a normalized value of a duration parameter corresponding to the target text, a normalized value of the number of target word objects, and an average priority of the target word objects.

[0052] In a specific embodiment, the score impact indicator set corresponding to the target text is obtained by the following steps:

[0053] S201, according to several target texts corresponding to a given word object, normalize the duration parameter corresponding to each target text to obtain a normalized value of the duration parameter; wherein, the duration parameter corresponding to the target text refers to the duration between the publication time of the target text itself and the publication time of the first text; the first text refers to the target text with the earliest publication time; it can be understood as: normalizing the duration parameter corresponding to each target text to a value between 0 and 1.

[0054] S202, according to a number of target texts corresponding to a given word object, normalize the number of target word objects corresponding to each target text to obtain a normalized value of the number of target word objects; this can be understood as: normalizing the number of target word objects corresponding to each target text to a value between 0 and 1.

[0055] S203, for any target text corresponding to a given word object, calculate the TF-IDF value corresponding to each target word object in the target text, and take the average of several TF-IDF values ​​as the average priority of the target word object to obtain a score influence indicator set corresponding to the target text; those skilled in the art know the specific process of calculating the TF-IDF value, which will not be repeated here.

[0056] As mentioned above, since these three score influencing indicators are all related to the given technology, these three indicators can reflect the correlation between the corresponding target text and the given technology, that is, reflect the score of the target text, which is conducive to obtaining the constant value in the subsequent preset life cycle indicator model to obtain the final technology maturity prediction model.

[0057] S300, classifying a plurality of target texts according to text types to obtain a plurality of target text sets.

[0058] S400, performing unsupervised clustering on each target text set to obtain a number of key texts, so as to obtain a score influence indicator set corresponding to each key text; the key text refers to the target text corresponding to the center of any target text cluster obtained by unsupervised clustering.

[0059] As mentioned above, by classifying the target texts so that the target texts in the same set are of the same type, the accuracy and reliability of clustering can be improved. By unsupervised clustering and screening out the center of each target text cluster as the screened key text, the number of texts used can be reduced, greatly reducing the subsequent computational complexity.

[0060] S500: taking the normalized value of the duration parameter in the score impact index set corresponding to the i-th key text, the normalized value of the number of target word objects, the average priority of the target word objects, and the preset score corresponding to the i-th key text as x i1 、x i2 、x i3 and Z i Substitute into the preset life cycle indicator model Z i =ax i1 +bx i2 +cx i3 +d, the to-be-processed lifecycle indicator model corresponding to the i-th key text is obtained, where a, b, c, and d are constants.

[0061] S600, according to the pending life cycle indicator model corresponding to each key text, the values ​​of a, b, c, and d are obtained through a genetic algorithm to obtain a target life cycle indicator model, and the target life cycle indicator model is used as a technology maturity index; it can be understood as: using a genetic algorithm to find the optimal solution for the formulas corresponding to several pending life cycle indicator models to obtain the optimal values ​​of a, b, c, and d; those skilled in the art know the specific implementation method of the genetic algorithm for finding the optimal solution, and the solution process is a prior art and does not belong to the innovative part, so it will not be repeated here.

[0062] As mentioned above, since the normalized value of the duration parameter, the normalized value of the target word object, and the average priority of the target word object are all close features related to the text and the given technology, these three indicators are used as variables in the preset life cycle indicator model and a series of calculations are performed to obtain a reasonable technology maturity index corresponding to the given technology.

[0063] S700: Obtain a technology maturity prediction model corresponding to a given word object according to the technology maturity index and the pre-acquired target growth limit.

[0064] In a specific embodiment, the target growth limit is obtained through the following steps:

[0065] S701, according to the publication time of each target text, obtain a number of target text annual growth rates corresponding to the target texts; it can be understood as: calculating the annual text growth rate of the target text in units of years.

[0066] S702: taking the largest target text annual growth rate among several target text annual growth rates as the target growth limit.

[0067] Specifically, the formula corresponding to the technology maturity prediction model meets the following conditions:

[0068] F=L / (1+e -z ), where F is the technology maturity corresponding to a given word object, L is the maximum text growth rate obtained in advance, and z is the technology maturity index.

[0069] As mentioned above, on the basis of the existing technology maturity curve prediction model, the technology maturity index has been improved through the relevant data of a large number of target texts related to the technology, so that the technology maturity prediction model corresponding to any technology can be obtained, and the obtained prediction curve is more accurate and reliable, which is helpful to understand the development level of the technology, thereby providing a research basis for the subsequent development direction.

[0070] An embodiment of the present invention also provides a non-transitory computer-readable storage medium, which can be set in an electronic device to store at least one instruction or at least one program related to implementing a method in a method embodiment. The at least one instruction or the at least one program is loaded and executed by the processor to implement the method provided in the above embodiment.

[0071] An embodiment of the present invention further provides an electronic device, comprising a processor and the aforementioned non-transitory computer-readable storage medium.

[0072] An embodiment of the present invention further provides a computer program product, which includes program code. When the program product is run on an electronic device, the program code is used to enable the electronic device to execute the steps of the method according to various exemplary embodiments of the present invention described above in this specification.

[0073] Although some specific embodiments of the present invention have been described in detail by way of example, it will be appreciated by those skilled in the art that the above examples are intended to be illustrative only and not to limit the scope of the present invention. It will also be appreciated by those skilled in the art that various modifications may be made to the embodiments without departing from the scope and spirit of the present invention. The scope of the present invention is defined by the appended claims.

Claims

1. A method for obtaining a prediction model, characterized in that: The method comprises the following steps: According to a plurality of target word objects in the category to which the given word object belongs, a plurality of target texts corresponding to the given word object are acquired from a preset text library; the given word object is the name of any technology specified by the user; Obtaining a score impact indicator set corresponding to each target text; the score impact indicator set corresponding to the target text includes a normalized value of a duration parameter corresponding to the target text, a normalized value of the number of target word objects, and an average priority of the target word objects; Classify a number of target texts according to text types to obtain a number of target text sets; Perform unsupervised clustering on each target text set to obtain a number of key texts, so as to obtain a score influence indicator set corresponding to each key text; the key text refers to the target text corresponding to the center of any target text cluster obtained by unsupervised clustering; The normalized value of the duration parameter in the score impact index set corresponding to the i-th key text, the normalized value of the number of target word objects, the average priority of the target word objects, and the preset score corresponding to the i-th key text are taken as x i1 、x i2 、x i3 and Z i Substitute into the preset life cycle indicator model Z i =ax i1 +bx i2 +cx i3 +d, obtain the pending life cycle indicator model corresponding to the i-th key text, where a, b, c, and d are constants; According to the pending life cycle indicator model corresponding to each key text, the values ​​of a, b, c, and d are obtained by genetic algorithm to obtain the target life cycle indicator model, and the target life cycle indicator model is used as the technology maturity index; According to the technology maturity index and the pre-acquired target growth limit, a technology maturity prediction model corresponding to a given word object is obtained.

2. The method for obtaining a prediction model according to claim 1, characterized in that: Get the target text corresponding to a given word object through the following steps: According to a plurality of target word objects in the category to which a given word object belongs that are obtained in advance, a plurality of first texts to be processed corresponding to each target word object are obtained from a preset text library, and a plurality of target sentences are determined from each first text to be processed; the target sentence refers to a sentence including at least one target word object obtained from a target paragraph in any first text to be processed; For any first text to be processed, according to the quantity information of each target word object in each target sentence in the first text to be processed, obtaining the first text priority corresponding to the first text to be processed; Obtain a number of second texts to be processed corresponding to the first text to be processed, and obtain the second text priority corresponding to the first text to be processed according to a number of target word objects corresponding to each second text to be processed and the time between the publication time of the first text to be processed and the publication time of each second text to be processed; the second text to be processed refers to any other first text to be processed whose publication time is earlier than that of the first text to be processed; When the sum of the first text priority and the second text priority corresponding to the first text to be processed is greater than a preset priority threshold, the first text to be processed is used as the target text corresponding to the given word object.

3. The method for obtaining a prediction model according to claim 2, characterized in that: The first text priority corresponding to the first to-be-processed text is obtained by following the steps below: For any target sentence in the first to-be-processed text, the target word objects in the target sentence are divided into a plurality of target word groups; the target word group refers to any word group in which the number of target word objects screened out from the currently remaining target word objects is the largest and each screened out target word object has only one; the currently remaining target word objects refer to the remaining target word objects obtained after deleting all the screened out target word objects from the target word objects in the target sentence; For any target phrase corresponding to the target sentence, the first phrase priority corresponding to the target phrase is calculated according to the total number of target word objects in the target phrase and the target number corresponding to the target phrase; The target number corresponding to the target phrase refers to the number of phrases obtained from the target phrases corresponding to the target sentence that are identical to any target phrase; According to the same target phrase information in the target phrases corresponding to the target sentence, a plurality of first phrase priorities are deduplicated to obtain a plurality of second phrase priorities; Determining the sum of the priorities of the plurality of second phrases as the sentence priority corresponding to the target sentence; The sentence priorities corresponding to several target sentences in the first text to be processed are added together to obtain a first text priority corresponding to the first text to be processed.

4. The method for obtaining a prediction model according to claim 2, characterized in that: The second text priority corresponding to the first to-be-processed text is obtained by following the steps below: According to the time between the publication time of the first to-be-processed text and the publication time of each second to-be-processed text, each time is normalized to obtain a time proportion parameter corresponding to each second to-be-processed text; For any second text to be processed, according to a number of target word objects corresponding to the second text to be processed, the number of target word objects repeated between the second text to be processed and the first text to be processed is obtained; Determine the product of the time proportion parameter corresponding to the second text to be processed and the total number of target word objects repeated between the second text to be processed and the first text to be processed as the text weight of the second text to be processed; The ratio of the sum of the text weights of each second text to be processed to the number of the second texts to be processed is calculated, and the calculation result is used as the second text priority corresponding to the first text to be processed.

5. The method for obtaining a prediction model according to claim 1, characterized in that: Obtain the score impact indicator set corresponding to the target text through the following steps: According to a number of target texts corresponding to a given word object, a duration parameter corresponding to each target text is normalized to obtain a normalized value of the duration parameter; wherein the duration parameter corresponding to the target text refers to the duration between the publication time of the target text itself and the publication time of the first text; the first text refers to the target text with the earliest publication time; According to a number of target texts corresponding to a given word object, the number of target word objects corresponding to each target text is normalized to obtain a normalized value of the number of target word objects; For any target text corresponding to a given word object, the TF-IDF value corresponding to each target word object in the target text is calculated, and the average of several TF-IDF values ​​is used as the average priority of the target word object to obtain the score influence indicator set corresponding to the target text.

6. The method for obtaining a prediction model according to claim 1, characterized in that: Obtain the target growth limit by following the steps below: According to the publication time of each target text, several text annual growth rates corresponding to several target texts are obtained; The largest annual text growth rate among several annual text growth rates is taken as the target growth limit.

7. The method for obtaining a prediction model according to claim 1, characterized in that: The formula corresponding to the technology maturity prediction model meets the following conditions: F=L / (1+e -z ), where F is the technology maturity corresponding to a given word object, L is the maximum text growth rate obtained in advance, and z is the technology maturity index.

8. A non-transitory computer-readable storage medium, wherein at least one instruction or at least one program is stored in the storage medium, characterized in that: The at least one instruction or the at least one program is loaded and executed by the processor to implement the method for obtaining the prediction model as described in any one of claims 1-7.

9. An electronic device, characterized in that: The invention comprises a processor and the non-transitory computer-readable storage medium as claimed in claim 8.

Images