Philosophical Research on Enterprise Innovation Ecology Based on a Human–Computer Interaction Mental Model

Abstract

Enterprise innovation ecology plays a vital role in the operation and development of enterprises. It can enable enterprises to have benign and sustainable progress space in action. At the same time, it can also make the innovation ability of enterprises grow unprecedentedly, which is a crucial factor related to the survival and lifeline of enterprises. However, enterprise managers are often too immersed in concept innovation and structural optimization in the actual innovation and ecological construction of enterprises. Still, they do not pay attention to the problems in actual construction. This would cause the actual operation of the system to be restricted by key influencing factors such as environmental protection policies, making it difficult to achieve effective progress. In this regard, the purpose of this paper is to analyze and discuss the enterprise innovation ecological philosophy based on the human–computer interaction mental model. Therefore, this paper comprehensively considers four key factors that affect the construction of enterprise innovation ecology, namely, the promotion of enterprise technology, the competitiveness of enterprise market, the innovation characteristics of enterprises, and the environmental constraints of innovation ecology, and improves and perfects the innovation ecosystem from these factors. Finally, an efficient ecological model of enterprise innovation was designed, and the scientific construction was carried out through the human–computer interaction mental model combining human–computer interaction technology and cognitive psychology. At the same time, the model is applied to the actual enterprise operation, and the naive Bayesian algorithm is adopted to optimize the relevant performance of the algorithm model. The experimental results showed that the highest sample accuracy of the algorithm reaches 92.9%, which greatly improved the operation efficiency of enterprise innovation ecological construction.

1. Introduction

In the overall operating system of the enterprise, technological innovation, concept innovation, organizational innovation, and other innovative fields have always been the focus of enterprise managers and are striving for improvement. This is because innovation capability is related to the future development prospects of an enterprise and determines whether an enterprise can still have room for development and progress in the context of the ever-changing times. In this case, if an enterprise can stay at the forefront of innovation, it must focus on the ecological construction of innovation. This calls for firms to use ecological building as the benchmark in all areas, including resource consumption and enterprise environmental transformation, and for the innovation system to have the ecological features of sustainable recycling and green development. The ecological building of enterprise innovation frequently only progresses to the preliminary planning stage, and the design of the existing associated systems does not account for the actual issues with resource allocation. In order to solve the above problems, we can start from the important factors that affect the construction of enterprise innovation ecology, analyze the relationship between specific model construction and these factors, and combine the concept of innovation ecology with philosophical concepts. At the same time, in order to scientifically support innovative ecological construction, this paper will introduce human–computer interaction technology to solve professional problems in the actual environment operation.

The practical significance of this study: by breaking the bottleneck of development, the soft power of enterprises can be improved, so as to achieve a new stage of leapfrog development and surpass the world’s advanced enterprises. The theoretical significance of this study is to study the innovation of corporate culture in order to enrich the existing theory of innovation methods.

Aiming at the problems existing in the ecological construction of innovation of existing enterprises, this paper introduces human–computer interaction technology, combines relevant algorithms to implement scientific application strategies, and uses the Naive Bayesian algorithm to optimize performance. The innovations of this paper are as follows: (1) In order to better analyze the user’s psychological cognition and feedback when using the system, this paper introduces concepts related to cognitive psychology and combines human–computer interaction technology to create a human–computer interaction mental model. (2) The naive Bayesian algorithm under machine learning is used to optimize the performance of the algorithm model, which can greatly improve classification accuracy.

2. Literature Review

The ecologicalization of enterprise innovation is related to the survival state and future development of enterprises, so it has attracted many scholars in the field of enterprise management to conduct in-depth research in this direction. Among them, Yin, combined with the actual economic situation of enterprises, analyzed the relationship between enterprise innovation ability and ecological construction and proposed ideal improvement measures. He explored the feasibility of the proposed improvements from multiple impact levels [1]. Jianping expounded on the development status of port logistics enterprises and the current innovation problems they are facing. Aiming at the above problems, it analyzed the defects of the operation concept of this type of enterprise and proposed a new innovative ecological concept to solve the problem [2]. Sun reviewed the impact of past policy implementation on enterprises and analyzed the effect of different factors on the ecologicalization of enterprise innovation. On this basis, he proposed the PSM-DID model to evaluate the impact of different factors on enterprise innovation and finally proved the effect of the model through experiments [3]. Huynh’s research pointed out that enterprises should pay attention to their living environment in the initial stage of entrepreneurship. Through the analysis of multiple entrepreneurial teams, the enterprise’s innovation ability can indirectly image the enterprise’s performance [4]. Enterprise innovation is a problem that must be faced in enterprise management. The above research indicates that enterprise innovation is in essence an ecological operation. In the enterprise innovation ecological model, the innovation advantages of other companies can be combined to maximize the effect of enterprise innovation, but there is a lack of accurate data analysis on enterprise innovation ecology using intelligent technology.

Human–computer interaction technology can be used to solve problems that have not been scientifically supported in the above research. Human–computer interaction technology has been an emerging multi-field cross-research topic in recent years, and many scholars have tried different research directions to tap the research potential of this technology. Among them, Ciasullo’s research shows that introducing artificial intelligence technology into the ecological analysis of enterprise innovation can effectively improve the intelligent analysis ability of data and promote the intelligent management of enterprise innovation [5]. Lan analyzed the current students’ psychological reflection and mental characteristics when they play sports. He proposed a training framework based on human–computer interaction technology to improve the psychological state of students given the psychological problems encountered by students in sports and finally verified the feasibility of the framework through simulation experiments [6]. Wenjuan expounded on the systematic scheme of common physical education teaching mode. For the problem of physical education teaching for students with movement disorders, which is challenging to deal with in this model, human–computer interaction technology is adopted to construct a physical education teaching communication model. He discussed the characteristics and advantages of the model and finally compared the model with the common physical education interactive interface [7]. Adem and his team reviewed in detail the impact and harm caused by the rapid spread of COVID-19 in various fields in recent years. He analyzed the changes in the learning system of the education industry due to COVID-19. Finally, he proposed the technical standards of human–computer interaction to analyze and evaluate the teaching quality of the education platform [8]. Yang expounded on the function and structure of the existing navigation system using human–computer interaction technology. He analyzed the shortcomings of these systems in actual use and then adopted a solution combining deep learning algorithms and human–computer interaction technology to improve the positioning capability of the navigation system [9]. Maestri listed several popular auditory analysis tools and compared the features and advantages of each device. Then, the benefits of the above tools were absorbed, the theoretical model of human–computer interaction technology was introduced to create a comprehensive auditory analysis instrument, and the functional advantages of human–computer interaction technology from example verification were summarized [10]. The above research on human–computer interaction technology broadened the application field of this technology and provided innovative ideas for developing this technology. The above studies do not deeply consider the user’s psychological and emotional interaction, and the important performance of the algorithm model has not reached a high level.

3. Construction of Enterprise Innovation Ecological Model Based on a Human–Computer Interaction Mental Model

3.1. Construction of Enterprise Innovation Ecological Model under Philosophical Thinking

The ecologicalization of enterprise innovation is the key factor for enterprises’ long-term sustainable development, including technological innovation, concept innovation, etc. Only when an enterprise is entirely innovative in all aspects can it actively overcome the difficulties and obstacles encountered in the development of the enterprise. Ecological construction can ensure that the enterprise can achieve the best resource consumption and build the enterprise into an ecological platform with a virtuous circle and green development [11,12]. Since many fundamental ideas in the study of enterprise innovation ecology cannot be explained by conventional ways of thinking, this paper opts to conduct philosophical research on enterprise innovation ecology using ideas from the philosophy of technology and ecological philosophy. It establishes the framework for the following practical application section. First, the enterprise innovation ecology’s impacting elements are investigated. The final realization process is complex and challenging, and the enterprise innovation ecology is impacted and constrained at multiple stages. Therefore, it is necessary to sort out all the influencing factors and their relationships to prepare for the construction of the innovation ecological system. The influencing factors of enterprise innovation ecology are shown in Figure 1.

The relationship diagram of influencing factors in Figure 1 applies the construction concept of the diamond model to fully demonstrate the relationship between the four influencing factors and the factors. Among them, the four factors that greatly influence the construction of enterprise innovation ecology are the promotion of enterprise technology, the competitiveness of the enterprise market, the innovation characteristics of enterprises, and the environmental constraints of innovation ecology. They influence each other and produce corresponding effects, and obvious changes in one of them would definitely affect the development trend of the other. At the same time, these four influencing factors are also restricted by the local government and environmental protection measures, which promote the development of enterprise innovation ecology in a more favorable direction.

After clarifying the influencing factors of enterprise innovation ecology and the relationship between the factors, this paper starts from these four influencing factors. After gradually analyzing the different design angles of the ecological innovation process and the required functional configuration, the corresponding concept system of the enterprise innovation ecologicalization is proposed. This paper aims to comprehensively improve the innovative environmental performance of enterprises. The basic structure of the conceptual model is shown in Figure 2.

The conceptual system of enterprise innovation ecology designed in Figure 2 takes the analysis angle and functional performance of enterprise innovation ecology as the main body of the design. Among them, the three levels of resources, technology, and relationship in the analysis point of view have an essential influence on the internal and external integration ability of enterprise innovation ecologicalization, taking ecological and economic development into account. The ability to integrate the entire enterprise’s interests and ecological construction inside and outside the enterprise would ultimately determine the development direction and performance of enterprise innovation and ecologicalization.

After perfecting the concept design through the above expressions to determine the general development direction of enterprise innovation ecology, this paper integrates the overall structure of the enterprise and designs an enterprise innovation ecology model based on philosophical thinking. The model takes into account the role of four factors that affect innovation ecology and extract key improvement information from the conceptual system. The final designed innovative ecological model would comprehensively promote the development of enterprise innovation and ecological development from four aspects. The specific mode design is shown in Figure 3.

Figure 3 illustrates how the novel ecological model proposed in this work would carry out technical advancements in four areas: resource conservation, energy conservation, environmental management, and environmental beautification. In order to make the best use of everything without wasting anything, resource-saving technology is utilized to fairly allocate and utilize practical resources like water resources and office space resources. Energy-saving technology is used to achieve the purpose of sustainable use by means of efficient energy conversion for the energy to be consumed in the production process of enterprises, such as heat energy and electric energy. Environmental governance technology can be used to rationally plan the corporate environment and improve the utilization of corporate space. Landscaping technology reasonably increases the greening area of the enterprise and transforms the enterprise environment into a greener direction. The ecological model of enterprise innovation proposed in this paper integrates the high-efficiency technologies of these four aspects, and tries to highlight the characteristics of enterprise innovation and ecologicalization, and improve the business environment of enterprises.

3.2. Human–Computer Interaction Technology Based on Cognitive Psychology

This paper will present the final operation effect of the enterprise innovation ecological model through human–computer interaction technology [13]. In order to better capture the user’s psychological process in the process of use and understand the user’s behaviour to achieve a more realistic human–computer interaction effect, this paper introduces the knowledge of cognitive psychology and human–computer interaction technology to combine and construct a human–computer interaction technology. The psychological model of computer interaction is applied to the improvement of the ecological model of enterprise innovation [14,15]. Cognitive psychology is the most influential research direction, which can well predict the psychological and cognitive processes of users [16]. Figure 4 shows the general psychological, cognitive process analysis of cognitive psychology for human beings.

Figure 4 demonstrates how humans engage in the three main types of mental activity while analyzing their perceived reality and the outside world: perception, thinking, and cognition. The cognitive form, which develops from reasonable feedback and interaction between the two worlds and maintains the outcomes of the interaction, and the perception form, which acquires internal and external information by various sources, are among them. Processing the internal and external data gathered by the form of perception, the form of thinking then transmits it to the human form or to the form of cognition for feedback operation. The human cognitive process is gradually established through the continuous interaction and processing of information [17].

The human–computer interaction mental model established by combining cognitive psychology can grasp the changes of people’s psychological information in time and build a feasible research model with practical effects based on this. This paper divides the psychological model of human–computer interaction into various model forms to construct the ecological model of enterprise innovation. The model human–computer interaction technology combines the idea of recursion and makes the model construction into a cyclic feedback process, which can be divided into four steps [18,19]. The following is a detailed description of the cycle process:

1.

Input interaction

In this process, the prediction model is used to predict the operating state of the system, and the probability is as follows:

$$\overline{{\phi }_h}(j)=Q\{N_h(j)/X^{j-1}\}={\displaystyle \sum _i{q}_{ih}{\phi }_h(j-1)}$$

(1)

Among them, $N_h$ is the constructed model representation, $q_{ih}$ is the success probability of the transfer process, and at time j, $q_{ih}=Q\{N(j)=N_h/N(j-1)=N_i\}$ and $X^{j-1}$ are the test sorting sequences. The success probability of the mixing process is expressed as:

$$\begin{array}{l}{\phi }_{i/h}(j-1/j-1)=Q\{N_i(j-1)/N_h(j),X^{j-1}\}\\ =\frac{1}{\overline{{\phi }_h}}{q}_{ih}{\phi }_i(j-1),i,j=1,2,\dots ,s\end{array}$$

(2)

Filter initial state input:

$$\begin{array}{l}{\stackrel{\wedge }{Y}}_h^0(j-1/j-1)=E[Y(j-1)/N_h(j),X^{j-1}]\\ ={\displaystyle \sum _{i=1}^s{Y}_i(j-1/j-1){\phi }_{i/h}(j-1/j-1)}\end{array}$$

(3)

Filter initial covariance input:

$$\begin{array}{l}{Q}_h^0(j-1/j-1)={\displaystyle \sum _{i=1}^s{\phi }_{i/h}(j-1/j-1)\{Q_i(j-1/j-1)}\\ +[Y_i(j-1)-{\stackrel{\wedge }{Y}}_h^0(j-1/j-1)]{[Y_i(j-1){\stackrel{\wedge }{Y}}_h^0(j-1/j-1)]}^T\}\end{array}$$

(4)

2.

Parallel filtering

The representations of Formulas (3) and (4) are expressed as the transmission values of the filters corresponding to the construction model $N_h$, and each filter performs an optimal estimation of the system state in a parallel manner and obtains the corresponding state estimate ${\stackrel{\wedge }{Y}}_h(j/j)$ and the covariance matrix $Q_h(j/j)$.

3.

Real-time transformation of system model probability

After all system models successfully perform parallel filtering operations, the success probability ${\phi }_h(j)$ of all models is calculated and updated through the optimal function ${\mathrm{\Psi }}_h(j)$ of statistical model parameters. The results of ${\mathrm{\Psi }}_h(j)$ can be calculated from the residual value ${\tau }_h$ and residual covariance $U_h$ calculated in the parallel filtering process:

$$\begin{array}{l}{\mathrm{\Psi }}_h(j)=Q\{X(j)/N_h(j),X^{j-1}\}\\ =\frac{1}{{(2\pi )}^{m/2}|U_h(j){|}^{\frac{1}{2}}}\exp \{-\frac{1}{2}{\tau }_h^T(j)U_h^{-1}(j){\tau }_h(j)\}\end{array}$$

(5)

The final real-time transform value of ${\phi }_h(j)$ is calculated by the Bayes’ theorem:

$${\phi }_h(j)=Q\{N_h(j)/X^j\}=\frac{\overline{{\phi }_h}(j){\mathrm{\Psi }}_h(j)}{{\displaystyle \sum _{i=1}^s\overline{{\phi }_h}{\mathrm{\Psi }}_i(j)}}$$

(6)

4.

Output interaction

After calculating the probability of re-updating each model after obtaining the result, the ideal state estimation and covariance results are calculated by accumulating the state estimates of the filters according to the corresponding weight ratios [20]:

$$\stackrel{\wedge }{Y}(j/j)={\displaystyle \sum _{h=1}^s{\stackrel{\wedge }{Y}}_h}(j/j){\phi }_h(j)$$

(7)

$$\begin{array}{l}Q(j/j)={\displaystyle \sum _{h=1}^s{\phi }_h(j)\{Q_h(j/j)}+\\ [{\stackrel{\wedge }{Y}}_h(j/j)\stackrel{\wedge }{Y}(j/j)]\times {[{\stackrel{\wedge }{Y}}_h(j/j)\stackrel{\wedge }{Y}(j/j)]}^T\}\end{array}$$

(8)

3.3. Naive Bayesian Algorithm Optimization Scheme

Considering that the model constructed in this paper would be applied to large-scale enterprise systems, this paper would adopt the Naive Bayes algorithm to further optimize the performance of the algorithm model, such as classification accuracy. According to the classification characteristics of the Naive Bayes algorithm, the probability calculation process of classifying the sample $Y_s$ to be classified into the class $D_i$ can be expressed by the following formula:

$$Q(D_i|Y_s)=\frac{Q(D_i)Q(Y_s|D_i)}{Q(Y_s)}=\frac{Q(D_i)Q(k_1,k_2,\dots ,k_n|D_i)}{Q(k_1,k_2,\dots ,k_n)}$$

(9)

In Formula (9), $Q(k_1,k_2,\dots ,k_n)$ appears in the category as a constant, so it is not necessary to include it in the following formula derivation, Formula (10) simplifies the expression of Formula (9):

$$D(Y_s)={\arg }_{D_i\in D}\max Q(D_i)Q(k_1,k_2,\dots ,k_n|D_i)$$

(10)

In this algorithm, each feature item must not affect each other to ensure that each category would not be associated with other feature items [21], and satisfy the requirement that each condition is independent of the other. According to the above description, the following Formula can be deduced:

$$Q(k_1,k_2,\dots ,k_n|D_i)={\displaystyle \prod _{h=1}^{n}Q(k_h|D_i)}$$

(11)

Formula (11) can be used to improve Formula (10) [22] to obtain Formula (12):

$$D(Y_s)={\arg }_{D_i\in D}\max Q(D_i){\displaystyle \prod _{h=1}^{n}Q(k_h|D_i)}$$

(12)

The representation of a quantity with magnitude and orientation in the sample space that needs to be classified can be rewritten as:

$$Y_s=\{(k_1,u_1),(k_2,u_2),\dots ,(k_h,u_h),\dots ,(k_n,u_n)\}$$

(13)

In Formula (13), $u_h$ is the weight representation of the corresponding feature item $k_h$, that is, whether $k_h$ appears in the sample $Y_s$ that needs to be classified, $k_h$ has a corresponding limited range. If $u_h=0$, it means that the feature item has no relationship with $Y_s$; if $u_h=1$, it means that $k_h$ exists in the sample $Y_s$ that needs to be classified.

In this paper, the above algorithm is further improved, and the information tree strategy is introduced to optimize the classification effect of algorithm [23]. The final optimized Naive Bayes algorithm model is shown in Figure 5.

In the model structure shown in Figure 5, all sample attribute nodes $B_i$ are configured with corresponding weight nodes $U_i$, which can be used to describe the attribute relationship between samples. The following is a detailed description of the optimization algorithm.

First, the Laplace’s law is used to calculate the correlation probability [24], which is mainly used in this paper to calculate the frequency of occurrence of the category $D_h$ in the D set to which the category belongs, and finally the posterior probability $Q(d_h)$ can be estimated:

$$Q(D=d_h)=\frac{{\displaystyle \sum _{h=1}^{m}G(d_h)+1}}{tota{l}_M+tota{l}_D}$$

(14)

In Formula (14), ${\displaystyle \sum _{h=1}^{m}G(d_h)}$ is the sum of the frequency of occurrence of $d_h$ in the set containing the same kind of samples, $tota{l}_M$ is the sum of the number of samples used in the training process, and $tota{l}_D$ is the sum of the number of samples of the class set.

According to the law of large numbers, the total number of times the eigenvalue $b_i$ of the sample category $d_h$ is used in the sample is tested, and the conditional probability $Q(B=b_i|D=d_h)$ is calculated:

$$Q(B=b_i|D=d_h)=\frac{{\displaystyle \sum _{i=1,h=1}^{m}G(b_i,d_h)+1}}{{\displaystyle \sum _{h=1}^{m}G(d_h)+tota{l}_B}}$$

(15)

In Formula (15), ${\displaystyle \sum _{h=1}^{m}G(b_i,d_i})$ is the total number of times the feature value $b_i$ of the sample category $d_h$ is used in the sample during the sample training process, and $tota{l}_B$ is the total number of sample features.

The characteristic probability $Q(b_i)$ of the sample can be calculated by Formulas (14) and (15):

$$Q(b_i)={\displaystyle \sum _{h=1}^{m}Q(b_i|d_h)Q(d_h)={\displaystyle \sum _{h=1}^m\frac{{\displaystyle \sum _{i=1,h=1}^{m}G(b_i,d_h)+1}}{{\displaystyle \sum _{h=1}^{m}G(d_h)+tota{l}_B}}}}$$

(16)

According to the above formula, $Q(d_h)$, $Q(B=b_i|D=d_h)$ and $Q(b_i)$ can comprehensively test the probability that the category of the sample that needs to be classified would be re-updated after the result is obtained:

$$Q(D=b_h|B=b_i)=\frac{Q(d_h){\displaystyle \prod _{i=1}^{m}Q(b_i|d_h)}}{{\displaystyle \sum _{h=1}^n[Q(d_h){\displaystyle \prod _{i=1}^{n}Q(b_i|d_h)]}}}$$

(17)

According to Formula (17), all the posterior probabilities of the category $d_h(1\le h\le n)$ to which the feature $b_i$ belongs can be calculated, and then the final classification category $d_e$ of $b_i$ is determined by the maximum value of these probabilities, namely $Q(d_e|b_i)>Q(d_h|b_i)$:

$$\{\begin{array}{l}{Q}_{\max }={\arg }_{d\in D}\max Q(d_h|b_i)={\arg }_{d\in D}\max {\displaystyle \prod _{i=1}^{m}Q(b_i|d_h)Q(D_h)}\\ d_e=\{f:d_h\Rightarrow Q_{\max }|(b_i\in B,d_h\ne null)\}\end{array}$$

(18)

On this basis, given an M-dimensional training sample set $W=\{w_1,w_2,\dots ,w_m\}$, according to Formula (19), the ideal information that must have for the classification operation of this sample set can be calculated:

$$\{\begin{array}{l}{q}_h=\frac{{\displaystyle \sum _{h=1}^{n}F(D_h,W)+1}}{total(|W|)},W\ne NULL\\ L(W)=-{\displaystyle \sum _{h=1}^n{q}_h{\log }_2(q_h),0\le q_h\le 1}\end{array}$$

(19)

Among them, $q_h$ represents the prior probability that a specific sample $w_i(i\le m)$ is classified into the category $D_h(h\le n)$. According to the weight value $\left|W_i\right|/\left|W\right|$, the ideal information that attribute B belongs to W can be tested, and the information gain $L\_gain(B)$ can be tested according to the offset of the original information requirement $L(W)$ and the new information requirement $L(W^{\prime })$. The specific calculation formula is as follows [25]:

$$\{\begin{array}{l}L(W^{\prime })={\displaystyle \sum _{i=1}^m\frac{\left|W_i\right|}{\left|W\right|}L(W_i)}\\ L\_gain(B)=L(W)-L(W^{\prime })\end{array}$$

(20)

Then, select the sample with the largest information gain rate $L\_gain\_ratio(B)$ for split operation and divide all the information into the correct classification:

$$L\_gain\_ratio(B)=\frac{L\_gain(B)}{L\_split(W)}=\frac{L(W)-L(W^{\prime })}{-{\displaystyle \sum _{i=1}^m\frac{\left|W_i\right|}{\left|W\right|}\ast {\log }_2\frac{\left|W_i\right|}{\left|W\right|}}}$$

(21)

Among them, $L-split(B)$ represents the split information of the attribute $B_i$ of the sample $W_i(1\le i\le m)$.

4. Application of Human–Computer Interaction Mental Model to Construction of Enterprise Innovation Ecology

4.1. Application of the Enterprise Innovation Ecological Model

This paper adopts a variety of experimental forms of experimental simulation and questionnaire surveys to relevant people in order to carry out research on the realization effect of this model. This is done in order to fully explore the viability, effectiveness, and performance of the human–computer interaction mental model and naive Bayesian algorithm applied to the enterprise innovation ecological model proposed in this paper. The first step is to determine how much the four components in this model, which have a significant impact on the ecologicalization of enterprise innovation, influence one another. Five expert researchers with a focus on enterprise innovation ecology are asked to analyze the impact of the four components on this topic as part of this publication. The range of each researcher’s influence evaluation for each element is 0 to 10, with higher scores indicating greater influence in the researcher’s appraisal of the component. The final results of the four factors are shown in

Table 1. Influence survey on factors influencing enterprise innovation ecology.

Factors Researcher	R1	R2	R3	R4	R5	Average Rating
Technology push	9	10	9	9	10	9.4
Innovative ecological environment conditions	9	9	9	10	9	9.2
Market allocation	9	9	8	8	9	8.6
Innovative enterprise characteristics	8	9	8	7	9	8.2
Other factors	7	7	6	6	8	6.8

. R1–R5 refer to five professional researchers in the field of enterprise innovation ecology.

As can be seen from Table 1, the average scores of the four influencing factors are all over 8 points, of which the average score of the technology-driving factor is the highest, reaching 9.4 points, and the average score of the lowest enterprise innovation characteristic factor is also 8.2 points. It can be seen that the five professional researchers have given high evaluations of the impact of these four factors on the ecologicalization of enterprise innovation. In contrast, the average score of other factors combined is only 6.8 points, which means that no other factor can have a greater impact on innovation ecology than the above four factors.

This paper would also show the specific functions and design plans of the designed enterprise innovation ecological model to five researchers in detail, so as to explore the innovation, ecological degree, and philosophical performance of the model in the analysis of professional fields. It provides a professional idea for the improvement of subsequent mode-related features. Then, this paper initiated an evaluation of the degree of manifestation of the relevant characteristics of the proposed model to five researchers, and the evaluation results are as follows.

The data in

Table 2. Evaluation of relevant characteristics of the enterprise innovation ecological model.

Factors Researcher	R1	R2	R3	R4	R5	Average Rating
Innovative	9	9	9	9	9	9
Degree of ecological	10	9	10	10	10	9.8
Philosophical thinking	8	8	9	9	8	8.4
Overall binding effect	9	9	8	9	9	8.8

show that the five researchers expressed high affirmation for the innovation, ecological degree, and philosophical thinking of the enterprise innovation ecological model proposed in this paper. Among them, the average score of the ecological degree reflected by the model is 9.8 points, and four researchers have given full marks. This paper has significantly improved the ecologicalization of enterprise innovation, which reflects this model’s good ecological design effect.

In addition, the ecological model of enterprise innovation proposed in this paper is optimized for the scope covered by the above four influencing factors. By combining four improvement measures to lead enterprises to the path of innovation ecology, in order to test the fitting effect of this model on many factors, this paper selects relevant samples of this model for data measurement and uses professional modeling tools to conduct simulation experiments. In the experiment, multiple indicators at the level of fitting indicators in relative and absolute dimensions were selected to measure the degree of fit. The final fitting effect of this model is as follows:

From the relevant data in

Table 3. Fitting effect data of the enterprise innovation ecological model.

Fit Metrics		Value	Fit to an Excellent Standard
Relative fit metrics	TLI	0.936	Value > 0.9
	CFI	0.918	Value > 0.9
	NFI	0.904	Values close to 0.9
Absolute fit metrics	RMSEA	0.049	Value < 0.08
	GFI	0.933	Value > 0.9
	CMIN	136.015	Value(CMIN)/Value(DF) < 2
	DF	91

, it can be seen that the index data in the model proposed in this paper meet the requirements of the fitting effect of all indicators. Among them, the Tucker–Lewis index (TLI), comparative fit index (CFI), and goodness-of-fit index (GFI) indicators can only reflect the excellent fitting effect when the index value is above 0.9. The values of these three indicators in this model are 0.936, 0.918, and 0.933, respectively, which obviously exceed the limit of excellent fitting effect. In the absolute fitting index, the fitting effect can only be judged as excellent when the ratio of CMIN and DF data is less than 2. The CMIN value and DF value of this mode are 136.015 and 91, respectively, and the ratio between the two not only meets the criterion but also greatly exceeds the criterion. From the above description, it can be seen that the overall fitting effect of the enterprise innovation ecological model proposed in this paper is in line with the excellent standard, and the design effect is ideal.

This paper conducts a correlation exploration experiment on the important influence indicators indicating the degree of enterprise innovation and ecologicalization. Through the actual experimental results, the degree of reflection of each impact index in the enterprise innovation ecological model constructed in this paper is analyzed and summarized. The load of the final innovation ecological index in this model is shown in

Table 4. Composition of innovation and ecological indicators in the model.

Number	Indicator	Indicator Innovative Component	Indicator Ecological Component
E1	Reduce resource consumption	0.211	0.795
E2	Reduce energy consumption	0.185	0.822
E3	Alternative to renewable energy	0.105	0.744
E4	Safer material alternative	0.139	0.751
E5	Reduce pollution	0.221	0.790
I1	Corporate image	0.836	0.277
I2	The overall competitiveness of the enterprise	0.912	0.115
I3	Market share	0.878	0.134
I4	Research and development investment	0.825	0.150
I5	Development trend	0.694	0.114

. E1 to E5 represent ecological indicators, and I1 to I5 represent innovative indicators.

Table 4 shows that, in the reflected component data of the ecological indicators in the model proposed in this paper, the ecological load components of the five ecological indicators in this model are all above 0.74. Among them, the ecological component reflected in the reduction of energy consumption index is the highest, with a value of 0.822. This also shows that this indicator has the highest degree of completion in the actual operation of the model. In the innovative load component data of innovative indicators, except for the development trend indicator, it is only 0.694. The innovative components of the remaining four indicators are all above 0.82, and the overall competitiveness of enterprises has reached the highest value of 0.912. Therefore, this model should pay more attention to the overall investment in development trend indicators and strive to reflect more fully the innovative ecological characteristics of this model.

The last section of this work also applies the enterprise innovation ecological model to the actual operation of an enterprise, choosing an outdated business with a weak innovation ecology as the experimental subject. This essay fosters the enterprise’s sustainable development while also enhancing the enterprise’s general operational environment and operating philosophy. Six months after the model started functioning, a satisfaction survey was given to the company’s employees. The objects of investigation are the new model proposed in this paper and the original model used by the enterprise before the improvement. A total of 202 employees of the company were selected for questionnaire survey, and 198 valid questionnaires were recovered. In the questionnaire survey, each employee can score two modes, respectively, and the scoring range is 0 to 10. The scoring results are shown in Figure 6.

It can be clearly seen from Figure 6 that the 198 employees of the enterprise are more satisfied with the newly-run enterprise innovation ecological model. Among them, there are 101 employees with a full score, 45, 26, and 24 employees with a score of 9, 8, and 7, respectively, and 0 employees with a score below 6. In contrast, 18 employees in the original model rated it from 0 to 5, and only 3 employees were rated full.

4.2. Application of the Naive Bayes Algorithm

In order to increase the classification accuracy of the final operational model’s algorithm and the performance of the enterprise innovation ecology model built on the mental model of human–computer interaction, this study suggests a naive Bayesian algorithm. This paper will conduct experimental analysis on the key algorithmic performance, such as time consumption, classification accuracy, and control error, in order to demonstrate the actual operation effect of the model that is ultimately proposed in this paper as well as the benefits of the naive Bayesian algorithm used in data classification. The relevant samples of the suggested model and other typical samples for evaluating the effectiveness of the algorithm are chosen as the experimental objects in this study. This paper tests the time consumption of sample processing of three mainstream classification algorithms, naive Bayes algorithm, decision tree algorithm, and the ID3 algorithm. The test data are shown in Figure 7.

As can be seen from Figure 7, the time consumption of the Naive Bayes algorithm used in this paper is the most ideal in most samples except the Zoo sample. Among them, in the relevant sample time consumption test of the mode proposed in this paper, the algorithm only takes 66 ms, while the other two algorithms take 101 ms and 86 ms, respectively. It can be seen that the Naive Bayes algorithm occupies an absolute advantage in time consumption.

Since the enterprise innovation ecological model proposed in this paper is applied in a huge system of the size of an enterprise, it has high requirements for time consumption. Considering that the time consumption performance of the Naive Bayes algorithm and ID3 algorithm in Figure 7 is better, this paper selects these two algorithms for further key performance comparison. This paper firstly selects relevant samples of the enterprise innovation ecological model for 2000 batches of mean square error evaluation. In order to strengthen the credibility of the experiment, this paper conducts two control experiments on the two algorithms under the same variables, and the error results are shown in Figure 8. Figure 8a is the first control experiment, and Figure 8b is the second control experiment.

In the error comparison of the two algorithms in Figure 8, the Naive Bayes algorithm used in this paper has better error control performance in both experiments. Among them, the algorithm can control the error below 0.23 after the training batch is more than 800 times, and the minimum value of the mean square error reaches 0.1, which is significantly better than the minimum value of the ID3 algorithm of 0.38.

In this paper, two control experiments were also conducted on the accuracy of the two algorithms in the same environment. The sample size was increased to 2500 on the original basis to improve the scientificity of the experiment. The accuracy comparison is as follows:

In Figure 9, the algorithm accuracy of the Naive Bayes algorithm used in this paper stabilizes at more than 89% after 2000 samples. Among them, the highest value of the accuracy is 92.9%, and the accuracy of the algorithm is higher than that of the ID3 algorithm in all sample testing stages, and the highest value of the ID3 algorithm is only 82.6%. It can be seen that the algorithm selected in this paper has huge advantages in terms of time consumption, error control, and accuracy improvement.

5. Conclusions

How to build an enterprise innovation ecological construction system that takes into account both efficiency and actual operation quality has always been a key topic of concern to many enterprise managers. This paper analyzes the advantages and disadvantages of the existing system based on the actual situation. According to the key factors affecting the ecological construction of enterprise innovation, a related concept system is established, and on this basis, combined with philosophical concepts, an efficient and feasible mode of ecological construction of enterprise innovation is constructed. Human–computer interaction technology is used to optimize the system design of this model. At the same time, in order to better meet the psychological needs of users, this paper combines this technology with cognitive psychology to construct an original human–computer interaction mental model. The core algorithm of the enterprise innovation ecological model based on the human–computer interaction mental model is the naive Bayesian algorithm, which can accurately analyze the actual enterprise innovation ecological data, and has excellent applications in time consumption, classification accuracy and control error. The naive Bayesian method used in this paper has achieved an accuracy of 89% after 2000 times of sampling. The results show that this method has great advantages in time consumption, error control, and accuracy. Based on this model, this paper proposes to introduce the Naive Bayes algorithm into the performance optimization of the algorithm model, which greatly optimizes the accuracy and error control performance of the enterprise innovation ecological model. However, due to the limitations of time and technology, this paper did not elaborate on the problems in the construction of the human–computer interaction mental model, which will be further discussed in the future.