Natural language processing scoring method, system and equipment based on focus adjustment theory and medium

Abstract

The invention discloses a natural language processing scoring method, system and device based on a focus adjustment theory and a medium, and relates to the technical field of natural language processing. The method comprises the following steps: constructing a corpus based on a focus adjustment theory; the corpus comprises entrepreneurship information and corresponding classification labels; constructing an improved SnwoNLP model on the basis of the corpus, and performing model parameter optimization by utilizing an Optuna framework to obtain a RegFousNLP model; the improved SnwoNLP model is constructed based on an analysis function of a promotion tendency and a defense tendency of an adjustment focus theory; calculating a target natural language by using the RegFousNLP model to obtain two classification numerical values; the two classification values comprise a promotion tendency value and a defense tendency value; and performing comprehensive calculation on the two classification values to obtain a final adjustment focus index. According to the invention, the efficiency and effect of entrepreneurship communication can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of natural language processing, and in particular to a natural language processing scoring method, system, device and medium based on regulatory focus theory. Background Art

[0002] In recent years, with the rapid development of Internet technology and mobile communication technology, emerging business models such as the sharing economy, e-commerce and artificial intelligence have been rapidly expanded. However, market competition is becoming increasingly fierce, and the challenges faced by entrepreneurs are also increasing. In such an environment, the communication strategy of entrepreneurs has become one of the key factors for their success. However, due to the diversity of cultural and economic backgrounds, how to adjust language to meet the expectations of different investors and consumers has become a problem that entrepreneurs need to solve urgently.

[0003] Regulatory focus theory provides a powerful tool for understanding and improving entrepreneurs' communication strategies, but it still has the following defects and shortcomings:

[0004] Limitations of general-purpose natural language processing tools: Natural language processing (NLP) tools currently on the market are mainly designed for a wide range of language understanding and generation tasks, such as sentiment analysis, text summarization, machine translation, etc. These tools perform well when processing general texts, but when applied to specific areas, such as entrepreneurial communication and business strategy analysis, they often lack the necessary customization and depth. For example, these tools may not be able to accurately identify and interpret complex business terms or industry-specific expressions in business plans, and it is difficult to conduct in-depth analysis of entrepreneurs' language styles to identify their regulatory focus tendencies.

[0005] Insufficient application of psychological motivation theory in NLP: Although regulatory focus theory has been widely studied in the fields of psychology and management, and its impact on individual behavior and decision-making processes has been confirmed, there are relatively few examples of effectively combining this theory with technology in the field of natural language processing. Existing NLP systems rarely consider the impact of users' psychological motivations and goal pursuit strategies on language use, which limits their application in business strategy analysis and entrepreneur guidance.

[0006] The personalized needs of entrepreneurs' communication strategies cannot be met: In a rapidly changing market environment, especially in a culturally diverse and rapidly developing country like China, entrepreneurs' communication needs are highly personalized and culturally sensitive. Existing NLP tools usually lack a deep understanding of regional cultural differences, industry-specific contexts, and personal communication styles. Therefore, they may not provide sufficient support for entrepreneurs to help them adjust their communication strategies to meet the expectations of different cultures and business environments.

[0007] Insufficient evaluation and training tools for entrepreneurial support institutions: Incubators, accelerators and other entrepreneurial support institutions lack tools that can accurately analyze and improve the communication effectiveness of entrepreneurs when evaluating and training resident companies. Existing technologies often cannot provide sufficient quantitative data and qualitative feedback to guide entrepreneurs on how to adjust their presentations and business documents according to their regulatory focus and market needs. Summary of the invention

[0008] The purpose of the present invention is to provide a natural language processing scoring method, system, device and medium based on regulatory focus theory, which can improve the efficiency and effectiveness of entrepreneurial communication.

[0009] To achieve the above object, the present invention provides the following solutions:

[0010] A natural language processing scoring method based on regulatory focus theory, comprising:

[0011] Constructing a corpus based on the regulatory focus theory; the corpus includes entrepreneurial information and corresponding classification labels; wherein the entrepreneurial information includes the entrepreneur's speech content in entrepreneurial programs, entrepreneurial interviews, and entrepreneurial documents;

[0012] An improved SnwoNLP model is constructed based on the corpus, and the model parameters are optimized using the Optuna framework to obtain a RegFocusNLP model; the improved SnwoNLP model is constructed based on the analysis function of the promotion tendency and the defense tendency of the regulatory focus theory;

[0013] Utilizing the RegFocusNLP model to calculate the target natural language, and obtaining two-category values; the two-category values ​​include a promotion tendency value and a defense tendency value;

[0014] The two classification values ​​are comprehensively calculated to obtain the final adjustment focus index.

[0015] Optionally, the corpus is constructed as follows:

[0016] The speech contents of entrepreneurs in entrepreneurial programs, entrepreneurial interviews, and entrepreneurial documents are sorted out, and labeled and classified according to the regulatory focus dictionary of regulatory focus theory. The categories of sentences are marked, and the promotion category is marked as 1 and the defense category is marked as 0. The punctuation marks of the collected text content are deleted, and the collected text is segmented using Jieba word segmenter, and the stop words in the text are removed to complete the construction of the corpus.

[0017] Optionally, the improved SnwoNLP model is constructed as follows:

[0018] The sentiment analysis function in the traditional SnwoNLP model was modified into an analysis function of promotion tendency and defense tendency based on regulatory focus theory, and the model was trained according to the corpus, Optuna framework, model training formula and bag-of-words model.

[0019] Optionally, the Optuna framework uses a TPE algorithm for parameter optimization.

[0020] Optionally, the two classification values ​​are comprehensively calculated to obtain a final adjustment focus index, specifically including:

[0021] Based on the promotion tendency value and the defense tendency value, a preliminary value of the regulatory focus index is obtained using a normalized interpolation method, and the preliminary value is smoothed using a hyperbolic tangent function tanh to obtain a final regulatory focus index.

[0022] The present invention also provides a natural language processing scoring system based on the regulatory focus theory, comprising:

[0023] A corpus construction unit, used to construct a corpus based on the regulatory focus theory; the corpus includes entrepreneurial information and corresponding classification labels; wherein the entrepreneurial information includes the entrepreneur's speech content in entrepreneurial programs, entrepreneurial interviews, and entrepreneurial documents;

[0024] A model building unit, used to build an improved SnwoNLP model based on the corpus, and optimize the model parameters using the Optuna framework to obtain a RegFocusNLP model; the improved SnwoNLP model is built based on the analysis function of the promotion tendency and the defense tendency of the regulatory focus theory;

[0025] A text classification unit, used to calculate the target natural language using the RegFocusNLP model to obtain two classification values; the two classification values ​​include a promotion tendency value and a defense tendency value;

[0026] The scoring calculation unit is used to perform comprehensive calculation on the two classification values ​​to obtain a final adjustment focus index.

[0027] The present invention also provides an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the natural language processing scoring method based on the regulatory focus theory.

[0028] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the natural language processing scoring method based on the regulatory focus theory as described above.

[0029] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0030] The present invention discloses a natural language processing scoring method, system, device and medium based on the regulatory focus theory, the method includes constructing a corpus based on the regulatory focus theory; the corpus includes entrepreneurial information and corresponding classification labels; based on the corpus, an improved SnwoNLP model is constructed, and the model parameters are optimized using the Optuna framework to obtain a RegFocusNLP model; the improved SnwoNLP model is constructed based on the analysis function of the promotion tendency and defense tendency of the regulatory focus theory; the target natural language is calculated using the RegFocusNLP model to obtain two classification values; the two classification values ​​include a promotion tendency value and a defense tendency value; the two classification values ​​are comprehensively calculated to obtain a final regulatory focus index. The present invention can improve the efficiency and effect of entrepreneurial communication. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. 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 paying creative labor.

[0032] Figure 1 It is a flowchart of the natural language processing scoring method based on the regulatory focus theory of the present invention;

[0033] Figure 2 This is a schematic diagram of the scoring work in this embodiment;

[0034] Figure 3 It is a word cloud diagram of the promotion tendency dictionary in this embodiment;

[0035] Figure 4 This is a word cloud diagram of the defense tendency dictionary in this embodiment. DETAILED DESCRIPTION

[0036] 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 ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0037] The purpose of the present invention is to provide a natural language processing scoring method, system, device and medium based on regulatory focus theory, which can improve the efficiency and effectiveness of entrepreneurial communication.

[0038] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0039] like Figure 1 As shown, the present invention provides a natural language processing scoring method based on regulatory focus theory, comprising:

[0040] Step 100: constructing a corpus based on the regulatory focus theory; the corpus includes entrepreneurial information and corresponding classification labels; wherein the entrepreneurial information includes the entrepreneur's speech content in entrepreneurial programs, entrepreneurial interviews, and entrepreneurial documents;

[0041] Step 200: constructing an improved SnwoNLP model based on the corpus, and optimizing the model parameters using the Optuna framework to obtain a RegFocusNLP model; the improved SnwoNLP model is constructed based on the analysis function of the promotion tendency and the defense tendency of the regulatory focus theory;

[0042] Step 300: Calculate the target natural language using the RegFocusNLP model to obtain two-category values; the two-category values ​​include a promotion tendency value and a defense tendency value;

[0043] Step 400: Comprehensively calculate the two classification values ​​to obtain a final adjustment focus index.

[0044] Based on the above technical solution, the following is provided: Figure 2-Figure 4 The embodiment shown.

[0045] In this embodiment, based on the promotion hypothesis and prevention hypothesis of the regulatory focus theory, the speech content of entrepreneurs in entrepreneurial programs, entrepreneurial interviews, and entrepreneurial documents is manually sorted, and the label classification is performed according to the regulatory focus dictionary of the regulatory focus theory, and a corpus of the regulatory focus theory is established. The collection topics of entrepreneurial programs and entrepreneurial interviews are the words that entrepreneurs respond to the host's questions, and the words they tell when introducing the content of the entrepreneurial project, which are collected and sorted using the video-to-text technology; the collection topics of entrepreneurial documents are the words that entrepreneurs use to introduce their entrepreneurial projects, and the speech content of entrepreneurs in the meeting records in the entrepreneurial documents is collected, and the optical character recognition (OCR) based on deep learning technology (CRNN+CTC) is used to convert non-digital text into digital text; in the later stage, the summary data is manually cleaned, the categories of sentences are marked, the promotion category is marked as 1, the defense category is marked as 0, the punctuation of the collected text content is deleted, the collected text is segmented using the Jieba word segmenter, and the meaningless stop words such as conjunctions and adverbs that appear in large quantities are removed, and a corpus based on the regulatory focus theory is established.

[0046] SnowNLP is a Python library for processing Chinese texts. Its design was inspired by another text processing library, TextBlob. SnowNLP facilitates users to perform Chinese natural language processing tasks, including sentiment analysis, text summarization, word segmentation, etc. Since SnowNLP is not a natural language processing tool based on the regulatory focus theory, this embodiment replaces its corpus and modifies its functions. Specifically, based on the sentiment analysis function of SnowNLP, its sentiment analysis function is based on the naive Bayes classifier. This embodiment modifies its sentiment analysis function into an analysis function of the promotion tendency and defense tendency based on the regulatory focus theory. The specific model training formula is as follows:

[0047]

[0048] Where P(c|x) is the probability that a given text (x) belongs to category (c) (promotion or defense). P(x|c) is the probability that text (x) appears in category (c), P(c) is the prior probability that any text belongs to category (c), and P(x) is the probability that text (x) appears.

[0049] The feature representation usually uses the bag-of-words model, in which the importance of each word is adjusted by TF-IDF. TF-IDF is a combination of two statistics, term frequency (TF) and inverse document frequency (IDF), to evaluate the importance of words in the text.

[0050]

[0051]

[0052] TF-DF(t,d,D)=TF(t,d)×IDF(t,D)

[0053] Among them, f t，d is the frequency of word (t) appearing in document (d), and N is the total number of documents in the document set (D).

[0054] Optuna is an automated machine learning (AutoML) framework that is specifically designed to automate the hyperparameter optimization process of machine learning models. Its working principle is based on efficient search strategies such as Bayesian optimization, genetic algorithms, TPE (Tree-structured Parzen Estimator), etc. Through these strategies, Optuna is able to explore the optimal combination of hyperparameters to improve the performance of the model. Specifically, Optuna performs hyperparameter optimization by defining an optimization objective (objective function), that is, a metric (such as error rate, accuracy, etc.) that the user wants to minimize or maximize. The user defines this objective by writing a function, which includes the training and validation process of the model and returns a score that needs to be optimized.

[0055] Among them, Optuna mainly uses the TPE algorithm for parameter optimization. The algorithm uses the idea of ​​Bayesian optimization to predict the probability distribution of model performance under given hyperparameters by constructing a probability model. That is, the TPE algorithm divides the hyperparameter space into two parts: parameters with better performance and parameters with worse performance. These two parts of parameters are modeled separately to form two probability models (l(x)) and (g(x)), where (l(x)) represents the probability density function of the parameters that make the objective function perform better, and (g(x)) represents the probability density function of the parameters that make the objective function perform worse. When sampling new parameters, TPE will sample parameters from (l(x)) instead of uniformly sampling from the original or entire parameter space. The sampling strategy is based on the following optimization formula:

[0056]

[0057] Here x * is the next parameter point determined during the evaluation process, that is, the point that maximizes the ratio of (l(x)) to (g(x)) is selected as the next sampling point. After each experiment, (l(x)) and (g(x)) are updated according to the experimental results. This update process relies on the premise of updating the posterior probability distribution in Bayesian theory, that is, constantly adjusting the parameters of the probability model with new experimental results. Through such an iterative process, TPE can gradually narrow the search range and focus on those hyperparameters that are most likely to bring better performance, thereby achieving more efficient hyperparameter optimization.

[0058] By replacing the modified SnwoNLP model and using Optuna to automatically optimize its parameters, a natural language processing model based on the regulatory focus theory suitable for the entrepreneurial field is constructed, namely the RegFocusNLP model. Based on this model, the promotion tendency value and the defense tendency value are calculated, and the classification of natural language text is completed. The promotion tendency value and the defense tendency value are the possibilities of the RegFocusNLP model classifying it as a promotion tendency or a defense tendency. Based on the obtained promotion tendency value and defense tendency value, the normalized interpolation method is used to obtain the preliminary value of the regulatory focus index, and the obtained preliminary value is smoothed by the Sigmoid adjustment method, that is, the hyperbolic tangent function tanh is used, which will limit the result between -1 and 1 and provide a smoother output when the input value is small (that is, the promotion and defense scores are similar). The formula is expressed as:

[0059]

[0060] In general, by replacing the corpus of SnowNLP and modifying the classification tasks and classification targets, and further optimizing the naive Bayes classifier used by SnowNLP through Optuna parameter optimization, a more accurate natural language classification task is achieved, and a natural language processing model (RegFocusNLP) suitable for the regulatory focus theory is constructed. The promotion tendency value and the defense tendency value are comprehensively calculated, and the regulatory focus index is calculated based on the normalization method. The Sigmoid adjustment method is used to smooth the calculated value to obtain the final regulatory focus index, and finally a natural language processing scoring system RegFocusNLP System based on the regulatory focus theory is constructed. This patent not only improves the accuracy of natural language classification tasks, but also provides entrepreneurs with a new tool to help them better understand and adjust their business strategies and communication methods. In addition, this system also provides an effective evaluation and training tool for entrepreneurial support agencies, further enhancing their ability to support entrepreneurs.

[0061] Therefore, the present invention has the following beneficial effects:

[0062] (1) By combining regulatory focus theory with natural language processing technology, the present invention specifically classifies and scores promotional and defensive tendencies. Compared with traditional sentiment analysis tools (which usually only distinguish between positive and negative sentiments), the present system can provide more detailed and targeted analysis. This precise classification can help entrepreneurs better understand the impression their texts (such as business plans, investment presentations, etc.) may leave on investors or potential partners, so that they can make necessary adjustments to match their business goals and market needs.

[0063] (2) By accurately identifying and analyzing regulatory focus tendencies in entrepreneurship-related texts, the present invention provides a powerful decision-making support tool. Entrepreneurs can use this tool to optimize their communication strategies and business plans to ensure that they are consistent with the company's long-term goals and market positioning. In addition, entrepreneurship support organizations such as incubators and accelerators can also use this system to evaluate and improve the performance of their resident companies, thereby allocating resources and support more effectively.

[0064] (3) Considering that the expression of promotion focus and defense focus may differ in different cultural contexts, the application of this system can help entrepreneurs make effective language and strategy adjustments when preparing for international market promotion and cross-cultural communication. Such adjustments can not only reduce cultural misunderstandings, but also enhance the attractiveness and competitiveness of entrepreneurial projects in the global market. By sensitively addressing and reflecting the needs of different markets and cultures, entrepreneurs can more successfully expand internationally.

[0065] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0066] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only used to help understand the core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A natural language processing scoring method based on regulatory focus theory, characterized in that: include: Constructing a corpus based on the regulatory focus theory; the corpus includes entrepreneurial information and corresponding classification labels; wherein the entrepreneurial information includes the entrepreneur's speech content in entrepreneurial programs, entrepreneurial interviews, and entrepreneurial documents; An improved SnwoNLP model is constructed based on the corpus, and the model parameters are optimized using the Optuna framework to obtain a RegFocusNLP model; the improved SnwoNLP model is constructed based on the analysis function of the promotion tendency and the defense tendency of the regulatory focus theory; Utilizing the RegFocusNLP model to calculate the target natural language, and obtaining two-category values; the two-category values ​​include a promotion tendency value and a defense tendency value; The two classification values ​​are comprehensively calculated to obtain the final adjustment focus index.

2. The natural language processing scoring method based on regulatory focus theory according to claim 1 is characterized in that: The construction process of the corpus is as follows: The speech contents of entrepreneurs in entrepreneurial programs, entrepreneurial interviews, and entrepreneurial documents are sorted out, and labeled and classified according to the regulatory focus dictionary of regulatory focus theory. The categories of sentences are marked, and the promotion category is marked as 1 and the defense category is marked as 0. The punctuation marks of the collected text content are deleted, and the collected text is segmented using Jieba word segmenter, and the stop words in the text are removed to complete the construction of the corpus.

3. The natural language processing scoring method based on regulatory focus theory according to claim 1 is characterized in that: The construction method of the improved SnwoNLP model is: The sentiment analysis function in the traditional SnwoNLP model was modified into an analysis function of promotion tendency and defense tendency based on regulatory focus theory, and the model was trained according to the corpus, Optuna framework, model training formula and bag-of-words model.

4. The natural language processing scoring method based on regulatory focus theory according to claim 1 is characterized in that: The Optuna framework uses the TPE algorithm for parameter optimization.

5. The natural language processing scoring method based on regulatory focus theory according to claim 1, characterized in that: The two classification values ​​are comprehensively calculated to obtain the final adjustment focus index, which specifically includes: Based on the promotion tendency value and the defense tendency value, a preliminary value of the regulatory focus index is obtained using a normalized interpolation method, and the preliminary value is smoothed using a hyperbolic tangent function tanh to obtain a final regulatory focus index.

6. A natural language processing scoring system based on regulatory focus theory, characterized in that: include: A corpus construction unit, used to construct a corpus based on the regulatory focus theory; the corpus includes entrepreneurial information and corresponding classification labels; wherein the entrepreneurial information includes the entrepreneur's speech content in entrepreneurial programs, entrepreneurial interviews, and entrepreneurial documents; A model building unit, used to build an improved SnwoNLP model based on the corpus, and optimize the model parameters using the Optuna framework to obtain a RegFocusNLP model; the improved SnwoNLP model is built based on the analysis function of the promotion tendency and the defense tendency of the regulatory focus theory; A text classification unit, used to calculate the target natural language using the RegFocusNLP model to obtain two classification values; the two classification values ​​include a promotion tendency value and a defense tendency value; The scoring calculation unit is used to perform comprehensive calculation on the two classification values ​​to obtain a final adjustment focus index.

7. An electronic device, characterized in that: It includes a memory and a processor, the memory is used to store a computer program, and the processor runs the computer program to enable the electronic device to execute the natural language processing scoring method based on the regulatory focus theory according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that: It stores a computer program, which, when executed by a processor, implements the natural language processing scoring method based on the regulatory focus theory as described in any one of claims 1 to 5.

Images