Analysis of Inﬂuencing Factors of Consumers’ Online Pur-chase Intention of Fresh Agricultural Products Based on Online Review

. Fresh agricultural products e-commerce has become an important way of agricultural products consumption, but China’s fresh agricultural products e-commerce is still in its infancy. It is urgent to ﬁnd out the inﬂuencing factors of fresh agricultural products consumers’ online purchase intention for the development of fresh agricultural products e-commerce. This paper extracts the main factors a ﬀ ecting consumers’ online purchase intention of fresh agricultural products by text mining of 5958 online review data of Tmall Fresh Channel. On this basis, the multiple linear regression model is used to analyze the key factors a ﬀ ecting consumers’ intention to purchase fresh agricultural products online. The results show that the factors that consumers pay attention to mainly include product quality, service quality and perceived quality. In addition to appearance, price, product description reliability, distribution service quality and competitors, other factors have a signiﬁcant impact on consumers’ online purchase intention and the impact is positive. This study expands the breadth and depth of the research on the inﬂuencing factors of fresh agricultural products consumers’ online purchase intention, and also provides targeted suggestions for enterprises to improve products and services, and promotes the healthy development of fresh agricultural products e-commerce.


Introduction
In order to stimulate economic development and increase farmers' income, China's fresh agricultural products industry has gradually developed. The development of this industry can solve the problem of single agricultural production structure in China to a certain extent and become an important force in adjusting agricultural structure. In recent years, with the popularity of the Internet ( according to the statistical report on the development of the Internet released by CNNIC in 2022, the scale of Chinese users reached 1.032 billion, and the Internet penetration rate reached 73.0%) and the improvement of online payment, logistics facilities and other related conditions, the online shopping market has achieved rapid growth (the size of online shopping users reached 842 million as of December 2021, accounting for 81.6% of the overall Internet users). Fresh agricultural products e-commerce can shorten the circulation channels, reduce the circulation links and costs, which provides a new idea for solving the dilemma of fresh agricultural products circulation. With the development of rural e-commerce, online shopping of fresh agricultural products has gradually become an important way of agricultural products consumption. However, China' s fresh agricultural products e-commerce is still in its infancy. Problems such as imperfect system and small application scale have led to difficulties in the development of fresh agricultural products e-commerce. 2021, the scale of China's agricultural products network transactions accounted for only 3.2% of the total online sales of social consumer goods, and the network marketing of agricultural products seriously lags behind. In recent years, the COVID-19 has had a great impact on the production, processing, transportation, sales and other links of fresh agricultural products. The original sales channels have been blocked, and a large number of fresh agricultural products have been unsalable. The supply chain and distribution are facing challenges such as high cost and low profit.
Fresh e-commerce is advancing rapidly under the impetus of reality. It is urgent to find out the influencing factors of fresh agricultural products consumers' online purchase intention for the development of fresh e-commerce. In recent years, many scholars have explored this, and some studies have explored the impact of fresh product characteristics on online purchase intention [1][2][3][4][5][6], and some scholars have explored the relationship between consumer characteristics and online purchase intention from the perspective of consumer characteristics [3,7,8], and some studies focused on platform and service characteristics to explore consumers' online purchase intention [3,5,6,8], the influence of brand and corporate image on consumers' online purchase intention is also the key research direction of scholars [6,9,10]. Online reviews play an important role between merchants and consumers. A large number of studies have shown that online reviews are the key factors affecting consumers' online purchase decision. The existing research on online reviews mainly focused on the e-commerce scenarios of all categories or specific categories, and the research entry points focus on the quantity and quality of reviews [11][12][13] and the positive negative tendencies of comments [14][15][16].
In recent years, the academic research on the online purchase intention of fresh agricultural products had focused on the influencing factors themselves, while the research on the influencing factors of consumers' online purchase intention of fresh agricultural products based on online reviews is relatively rare. Therefore, from the perspective of consumers, based on the online review data of consumers on the e-commerce platform of fresh agricultural products, this study identified and extracted the factors affecting purchase intention in online reviews through the NLPIR. Questionnaire survey, variance analysis, reliability and validity test, regression analysis and other methods were used to analyze the influencing factors.
The significance of this study is to expand the breadth and depth of the research on the influencing factors of consumers' online purchase intention of fresh agricultural products, to analyze the mechanism of each influencing factor in multiple dimensions, and to form a theoretical system with reference significance through comprehensive vertical and horizontal analysis. At the same time, it helps relevant enterprises to understand the influencing factors of consumers' online purchase intention of fresh agricultural products, provides targeted suggestions for enterprises to improve products and services, promotes the healthy development of enterprises, and also promotes the development of agricultural modernization. This paper reviewed the relevant literature in the second section, identified the influencing factors in the third section, puted forward the research hypothesis in the fourth section, made an empirical analysis of the hypothesis in the fifth section, and finally puted forward the conclusions and suggestions in the sixth section.

Concept Definition
(1) Online reviews Product reviews published by consumers on e-commerce or third-party websites are mainly expressed in the form of star ratings (1 to 5 stars) and open-ended review texts [12]. Online reviews are the key reference information for consumers to measure whether to buy or not before they consume.
(2) Fresh agricultural product Fresh agricultural products, also known as fresh produce, mainly include farming, livestock, aquatic products and primary processed products. According to previous research experience and actual investigation, Zhang divided fresh agricultural products into five categories of unprocessed primary agricultural products: fresh fruits, fresh vegetables, seafood and aquatic products, raw meat and eggs [17].

Review of Related Studies
(1) Influencing factors of fresh agricultural products consumers' online purchase intention In previous studies, scholars mainly discussed the online purchase intention of fresh agricultural products from four perspectives: product characteristics, platform characteristics, consumer characteristics and service characteristics. There were also a few studies starting from brand and corporate image to explore the influencing factors of consumers' online purchase intention of fresh agricultural products.
In terms of product characteristics, some scholars had used the theory of consumption value to construct a theoretical model of green label purchase intention of agricultural products. Through research, it was found that functional value and environmental value play an intermediary role between green label and consumer purchase intention of green agricultural products [1]. Eunok thought that consumers' trust in the traceability of agricultural products will positively affect online purchase intention [2]. At the same time, a large number of studies had proved that product safety, product quality, and product freshness were also important factors affecting consumers' online purchase intention [3][4][5][6]. In terms of platform characteristics, studies had shown that website information richness significantly affected consumers' intention to purchase agricultural products e-commerce [3]. In Yang' s research, it was proved that platform characteristics had a significant positive impact on perceived ease of use and perceived usefulness, while perceived ease of use and perceived usefulness have a significant positive impact on purchase intention [5]. Coincidentally, Lin et al. also pointed out in the research that platform characteristics (information quality, system quality, service quality) significantly affected consumers' perceived utilitarian value and perceived hedonic value, and perceived value played a key mediating role in the impact of platform characteristics on consumers' continuous purchase intention [6]. In terms of consumer characteristics, most scholars discussed from the perspective of demographic characteristics. For example, Wang believed that the gender and education level of consumers' individual attributes had a significant positive impact on the intention of online purchase for fresh agricultural products, while age had a significant negative impact on online purchase intention [3]. Gender was also one of the factors that affect the intention to purchase fresh agricultural products online. Women were more willing to purchase than men [7]. Some scholars had studied consumer cognition as the starting point. Empirical studies had found that product category cognition and brand cognition had a significant positive impact on the online purchase intention of fresh agricultural products [8], and consumers' product knowledge would also significantly positively affect their purchase intention [7]. In terms of service characteristics, delivery speed, interactive services and additional services had a significant positive impact on online purchase intention of fresh agricultural products [3,8]. Enterprise and brand image were also important factors affecting consumers' online purchase intention. Lee et al. pointed out in the study that brand image has the greatest impact on consumers' purchase intention [14]. Some studies also believed that social influence and corporate image have a significant positive impact on consumers' purchase intention of fresh e-commerce platform [9,10].
(2) Consumers' online purchase intention based on online reviews A large number of studies had shown that online reviews are a key factor affecting consumers' online purchase intention [18,19], many scholars had studied the influence of online reviews on consumers' online purchase intention. At present, the research on online reviews at home and abroad mainly focused on the following two aspects: The first aspect mainly measured the three factors of review content, recipient characteristics and reviewers. The second aspect is mainly reflected in the positive or negative tendencies of online reviews.
Chevalier and Mayzlin studied the evaluation content of online books based on Amazon.com and Barnesandnoble.com, and found that the content of online reviews has a significant impact on consumer purchase intention [11]. At the same time, research showed that online reviews have a positive correlation with sales volume, and online reviews will promote sales. Subsequently, Mudambi and Schuff studied 1587 Amazon.com online reviews of six products, and found that the depth of reviews had a positive impact on consumers' intention to purchase online [12]. Park found that the quality and quantity of online reviews have a significant positive impact on consumers' online purchase intention [13]. At the same time, the professionalism and credibility of reviewers and the professionalism, convergence and involvement of recipients significantly affected consumers' online purchase intention [20]. Tran also verified the positive effect of online reviews on purchase intention through the perceived effectiveness of social media platforms and online trust [21].
In addition, the positive or negative tendencies of online reviews also had a significant impact on consumers' online purchase intention. Zhao et al. analyzed the influencing factors and mechanism of online purchase intention of agricultural products based on the perspective of reference effect, and found that positive online reviews will positively affect the cognition and attitude of consumption to agricultural products [15]. Lee et al. found that positive reviews will positively affect consumers' online purchase intention, and would guide consumers to recommend products to people they know, and vice versa. Recommendations to people around, and even negative guidance [14]. Therefore, it can be seen that the positive and negative tendencies of online reviews will have an important impact on consumers' online purchase intention. Positive reviews or products with high reviews scores generally increased consumers' positive attitudes and purchase intentions [16].

Online Review Data Collection
At present, there are many fresh agricultural product e-commerce platforms in China, but Tmall ranks first in terms of overall strength. Therefore, this study used mobile Tmall fresh agricultural product online reviews as the data source. Pavlou argued that partial reviews can contain the content of most reviews, and most users only focused on the content of the first page of text reviews [22]. Therefore, the study selected the top 60 products in the five major categories of fresh agricultural product in terms of sales as the collection object respectively, and obtained the first 20 reviews of each product, totaling 6000 data. Then the review information was strictly screened according to the text content, and the screening principles were: eliminating deliberately praising and malicious vilification reviews; eliminating reviews with suspicion of advertising and reviews that did not mention the content related to the purchased goods [17]. The final data of 5958 valid reviews were obtained.

Online Review Data Processing
The valid data were put into a text file to get more than 540,000 words of review content, and high-frequency words in the text were extracted using NLPIR, a Chinese word separation system of the Chinese Academy of Sciences. To avoid the influence of deactivated words (meaningless for research), they were filtered out by manual screening, and the other similar high-frequency words were processed for synonymy conversion and merging. After processing, 186 high-frequency words (word frequency ≤ 10) were initially extracted.

Influencing Factors Identification
The high-frequency words were further sorted and merged, and 14 influencing factors were extracted by conceptualizing and summarizing the high-frequency words: fresh, appearance, portion size, preservation of freshness, price, nutrition, Guaranteed -merchant service level, reliability -product description, distribution service, logistics speed, promotion, giveaways, competitors, cooking method. The influencing factors were downscaled according to the CCSI model [23] and SERVQUAL scale [24] and grouped into three major categories as shown in table 1.

Research Hypothesis
Whether it was traditional purchase or online purchase, product quality was always the core issue in the purchase process, and product quality would directly determine whether consumers take the purchase behavior or not. He et al. found that consumers' expectations of product quality and safety have a significant impact on the purchase intention of fresh agricultural product online [3]. According to Li, the ease of assessing the quality of fresh agricultural product positively moderates the relationship between consumers' perceived quality and satisfaction, which influenced consumers' purchase intention [25]. Therefore, this paper proposed the following hypothesis. Hypothesis 1. There is a positive relationship between product quality and consumers' online purchase intention of fresh agricultural products.
With the increasing popularity of online shopping behavior, consumers' requirements for online shopping service quality were increasing, and consumers always chose merchants very carefully. Online stores provided consumers with complete product information, pertinent purchase advice, and timely service responses, which help to improve consumers' shopping experience and thus stimulate their purchase intentions. According to Wu, the degree of detail of product description, consistency between product description and actual, delivery speed, and merchant service significantly affect consumers' online purchase intention of fresh agricultural product [8]. Therefore, the following hypothesis is proposed in this paper.
Hypothesis 2. There is a positive relationship between service quality and consumers' online purchase intention of fresh agricultural products.
Perceived value, refers to the subjective evaluation of the utility of a product or service after the customer perceives it. The good or bad perceived value not only directly affected whether consumers themselves repeat the purchase behavior, but also indirectly affects other consumers' purchase intention of the product online. Based on the SOR model, Zhu analyzed the mechanism of the role of perceived value and consumers' repurchase intention and found that perceived value is an important indicator that drives consumers' repurchase intention [26]. Therefore, this paper proposes the following hypothesis.
Hypothesis 3. Perceived value has a positive relationship with consumers' online purchase intention of fresh agricultural products. In this paper, Cronbacha' s Alpha coefficient was used to test the reliability level of the scale. Cronbacha' s Alpha coefficient above 0.70 indicates that the reliability level is high, and the newly formed factors have good consistency. This study analyzed the reliability of three latent variables of product quality, service quality and perceived value of online fresh agricultural products. It can be seen from table 2 that the Alpha values of each latent variable are all greater than 0.85, and the reliability level was very high, indicating that the measurement items show good consistency.

Validity Test
For convergent validity, this paper used three criteria proposed by Fornell and Larcker to test, that was, the load of all standardized factors was greater than 0.5 and reached a significant level (p < 0.05 or p < 0.01); composite reliability (CR) greater than 0.7; the mean extracted variance (AVE) was greater than 0.7 [27]. The results of convergence validity were shown in table 3. The factor loadings of all observed variables were greater than 0.5 and reached significant level (P < 0.001). The average extracted variance (AVE) of each latent variable exceeded 0.5. The combined reliability (CR) was higher than the recommended value of 0.7; the fitness index of the measurement model basically met the requirements of the recommended value of the structural equation model (X 2 /df < 5; GFI, AGFI, CFI > 0.9; RMSEA < 0.08), and the measurement model has good convergence validity.  For the discriminant validity, the Fornell-Larcker criterion was still used in this paper. It is considered that the average extraction variance square root ( √ AV E) of a latent variable was greater than the correlation coefficient between the latent variable and other latent variables, which indicates that the latent variables have discriminant validity [27]. As shown in table 4, except that the square root of the average extraction variance of perceived value was slightly smaller than the correlation coefficient between perceived value and product quality, the rest were in line with the standard, indicating that the model has good discriminant validity.

Correlation Analysis
As shown in table 5, there was a highly significant positive correlation between different product quality components; six product quality components were highly significantly positively correlated with the intention to purchase fresh agricultural products online.   As shown in table 6, there was a highly significant positive correlation between different service quality components; four service quality components were highly significantly positively correlated with the intention to purchase fresh agricultural products online.
As shown in table 7, there was a highly significant positive correlation between different perceived value components; four perceived value components had a highly significant positive correlation with the intention to purchase fresh agricultural products online.
The regression results showed that the goodness of fit test of the model is 0.926, and the model can better explain the consumers' intention to purchase fresh agricultural products online. In terms of product quality, the significance of fresh (regression coefficient is 0.147, P < 0.01), portion size (regression coefficient is 0.125, p < 0.01), preservation of freshness (regression coefficient is 0.149, p < 0.01) and nutrition (regression coefficient is 0.161, p < 0.01) is qualified, and the significance of appearance (regression coefficient is 0.037, p > 0.05) and price (regression coefficient is 0.019, p > 0.05) is unqualified. Hypothesis 1 was partially verified. In terms of service quality, the significance of guaranteed-merchant service level (regression coefficient 0.094, p < 0.01) and logistics speed (regression coefficient 0.130, p < 0.01) is qualified, and the significance of reliability-product description (regression coefficient 0.061, p > 0.05) and distribution service (regression coefficient 0.036, p > 0.05) was unqualified. Hypothesis 2 is partially verified. In terms of perceived value, the significance of promotion (regression coefficient is 0.091, p < 0.01), giveaways (regression coefficient is 0.059, p < 0.05), cooking method (regression coefficient is 0.076, p < 0.01) was qualified, and the significance of competitors (regression coefficient is 0.018, p > 0.05) was unqualified. Hypothesis 3 is partially verified.

Research Conclusion
Based on the online review data of fresh agricultural products e-commerce platform, this study identified and extracted the influencing factors of online purchase intention. The influencing factors were empirically analyzed by means of questionnaire survey, reliability and validity test, correlation analysis and regression analysis. Through comprehensive vertical and horizontal analysis, the breadth and depth of relevant research were expanded, and a theoretical system with reference significance was formed. The specific conclusions of this study are as follows: (1) There was a positive relationship between product quality and purchase intention. First, the fresh of fresh agricultural products significantly and positively affects consumers' purchase intention of fresh agricultural products online. The fresh of fresh agricultural product in the online shopping process affected consumers' first impression of it and played an important role in consumers' evaluation of the overall purchasing experience. Second, the portion size of fresh agricultural product significantly and positively influenced consumers' purchase intention of fresh agricultural product online. The more accurate the portion of the product received by consumers, the more trustworthy it was, and the more willing consumers were to purchase fresh agricultural product online. Thirdly, the nutrition of fresh agricultural product significantly and positively influenced consumers' online purchase intention of fresh agricultural product. The richer the nutrition of fresh agricultural products, the easier it was to meet the nutritional needs of different consumers, which helps to increase consumers' purchase intention. Fourthly, the influence of the appearance and price of fresh agricultural products on online purchase intention had not been supported. According to the results of regression analysis, consumers paid attention to the factors affecting the value of fresh agricultural products such as freshness, preservation and nutrition when shopping fresh agricultural products online. The appearance did not affect the effect of meeting consumers' physiological needs, so its impact on consumers' online purchase intention was not significant. With the general improvement of living standards, people' s sensitivity to prices had decreased. In addition, the prices of agricultural products were under the control of the state, generally within the generally accepted level and the difference was not large.
(2) There was a positive relationship between service quality and online purchase intention. First, the service level of merchants significantly and positively affected consumers' purchase intention of fresh agricultural product online. Since consumers did not have specific access to the products in the online shopping process, the service level of merchants became one of the important factors for customers to choose the products. Second, logistics speed significantly and positively affected consumers' purchase intention of fresh agricultural products online. Fresh agricultural products were perishable, so the faster the logistics speed of fresh agricultural products, the stronger the consumers' online purchase intention of fresh agricultural products. Third, the reliability of product descriptions did not have a significant impact on consumers' purchase intention. This was because fresh agricultural products are difficult to standardize, and different consumers had different standards and expectations for the same fresh agricultural products. Fourth, the effect of commodity delivery service on consumers' purchase intention was not significant. It might be because with the continuous improvement and perfection of the logistics system, there was no significant difference in the commodity delivery services provided by major e-commerce platforms.
(3) There was a positive relationship between perceived value and purchase intention. Firstly, the cooking method of fresh agricultural products significantly and positively affected consumers' purchase intention, the more the fresh agricultural products provided by merchants match with the cooking method required by consumers, the higher consumers' purchase intention. Secondly, the promotional activities of fresh agricultural product merchants significantly and positively affected consumers' purchase intention, the promotion of fresh agricultural product would increase consumers' perceived value. Thirdly, fresh agricultural product with complimentary services significantly and positively affected consumers' purchase intention, , consumers were more willing to purchase produce with complimentary products than produce without complimentary products.

Management Implications
This study can help relevant enterprises to understand the influencing factors of consumers' online purchase intention of fresh agricultural products, provide targeted suggestions for enterprises to improve products and services, promote the healthy development of enterprises, and also promote the development of agricultural modernization. Specific recommendations are as follows: (1) Targeted guidance for online reviews. Fresh agricultural product e-merchants can provide corresponding evaluation options in the consumer's evaluation box, and set the corresponding level in each option for consumers to check, so as to provide more reference basis for other consumers. At the same time, merchants can encourage and support high-quality, multi-angle reviews to stimulate consumers' enthusiasm for online reviews.
(2) Optimize the product description framework of fresh agricultural products ecommerce platform. Fresh agricultural product merchants can make appropriate adjustments to the product description section: first, merchants can show the freshness of the product in the product reviews by various parties as much as possible, and can also regularly count the freshness, portion size and other factors affecting the rating in online reviews. Second, merchants can add short videos containing factors such as the origin of fresh agricultural products, processing and transportation processes, nutrition supply, and applicable cooking methods to the product interface to help consumers understand the products more intuitively.
(3) Improve the service level of merchants. Consumers' willingness to purchase fresh agricultural products online is closely related to their good shopping experience, and good shopping experience is the result of good merchant service. Therefore, in order to improve consumers' willingness to buy online, fresh agricultural product merchants should pay attention to the service level. Specifically, firstly, comprehensive service literacy training for employees can be carried out regularly. Good service attitude is the magic weapon to maintain customer relationship, and positive service attitude and handling are more likely to reduce negative comments. Second, enhance the interaction and communication with consumers, regular customer care, and thus enhance the goodwill of the company and willingness to buy. Third, regular promotional activities and appropriate gift-giving services for consumers to achieve the goal of attracting customer traffic and improving customer goodwill. In addition, in order to better reflect the level of service to different consumers, the fresh food e-commerce platform can also add the factor of "merchant service level" in the direction of guiding reviews for consumers to evaluate.

Research Insufficiency and Prospect
The shortcomings of this study are mainly reflected in two aspects: research methods and survey objects. In terms of research methods, in recent years, most of the studies on purchase intention used experimental methods to simulate real purchase scenarios, so that subjects can respond most instinctively to obtain relatively accurate behavioral results. This study mainly used the questionnaire survey method. The subjects responded to the subjective feelings of the questionnaire items, and the behavioral results might not be accurate enough. In terms of respondents, the main respondents of this study were college students and fourthtier urban residents. Their demand for online purchase of fresh agricultural products was not strong. Therefore, when filling out the questionnaire, they might rely on imagination rather than actual experience, resulting in the questionnaire results can not fully reflect their actual behavior.
In the future, the research on the purchase intention of fresh produce e-commerce can broaden the breadth and depth, and conduct in-depth research from different perspectives, so that relevant personnel can more deeply understand the influencing factors and mechanism of online purchase intention of fresh agricultural products. It can also discuss the emerging fresh produce e-commerce phenomenon (such as fresh live e-commerce), improve the existing ecommerce model and even contribute to building a new model.