Interface design method and device, electronic equipment and storage medium

Abstract

The invention discloses an interface design method and device, electronic equipment and a storage medium, and relates to the technical field of man-machine interaction, and the method comprises the steps: employing a crawler tool, and automatically collecting evaluation data for interface design on a network; dividing the collected evaluation data into positive evaluation data, negative evaluation data and neutral evaluation data based on text classification; identifying demand evaluation and negative evaluation in the evaluation data through a predefined syntactic structure, extracting key information in the evaluation data, and generating a visual report; based on the number and score of the user comments, setting a weighting coefficient of each function, and calculating a function priority; the feasibility of the design method is proved through an A / B test. According to the interface design method and device, the electronic equipment and the storage medium provided by the invention, the efficiency of user demand analysis is improved, a scientific basis is provided for interface design, and the market competitiveness of a product is enhanced.

Description

Technical Field

[0001] The present application relates to the field of human-computer interaction technology, and in particular to an interface design method, an interface design device, an electronic device and a storage medium. Background Art

[0002] In the process of interaction between people and machines, there is a level, which we call the interface. From a psychological point of view, the interface can be divided into two levels: sensory (visual, tactile, auditory, etc.) and emotional. User interface design is an important part of screen products. Interface design is a complex project involving different disciplines. Cognitive psychology, design, linguistics, etc. all play an important role here. The three principles of user interface design are: put the interface under the control of the user; reduce the user's memory burden; maintain the consistency of the interface.

[0003] However, interface design has not been taken seriously in the long process of software development. People who do interface design are also derogatorily called "artists". In fact, software interface design is an important selling point of products, just like industrial design in industrial products. A friendly and beautiful interface will bring people comfortable visual enjoyment, shorten the distance between people and the frame, and create selling points for businesses. Interface design is not just art painting. It needs to locate the user, the use environment, the use method and is designed for the end user. It is a purely scientific art design.

[0004] At present, the HMI design of automobile enterprises is guided by the subjective experience and aesthetics of designers, ignoring or having difficulty in capturing the actual needs of users. In order to capture user needs, designers often need multiple rounds of design attempts and subjective evaluations. The standard for testing an interface is neither the opinion of a project development team leader nor the result of a vote by project members, but the feelings of the end users. How to objectively and reasonably capture the subjective feelings of users and reflect them in interface design, and provide designers with scientific design methods, has become a key link in the development of interface design. Summary of the invention

[0005] The purpose of the present invention is to provide an interface design method, device, electronic device and storage medium, which not only improves the efficiency of user demand analysis, but also provides a scientific basis for interface design and enhances the market competitiveness of products.

[0006] The present invention provides the following scheme:

[0007] According to one aspect of the present invention, there is provided an interface design method, the interface design method comprising:

[0008] Use crawler tools to automatically collect evaluation data on interface design on the Internet;

[0009] Based on text classification, the collected evaluation data is divided into positive evaluation data, negative evaluation data and neutral evaluation data;

[0010] Through the predefined syntactic structure, identify the demand evaluation and negative evaluation in the evaluation data, extract the key information in the evaluation data, and generate a visual report;

[0011] Based on the number and ratings of user reviews, set the weighting coefficients for each function and calculate the function priority;

[0012] The feasibility of the design process was verified through A / B testing.

[0013] Optionally, use crawler tools to automatically collect evaluation data on interface design on the Internet, including:

[0014] Use crawler tools to automatically collect evaluation data;

[0015] Conduct preliminary identification and error correction on the collected batch reviews;

[0016] Among them, crawler tools include: BeautifulSoup, or Selenium.

[0017] Optionally, based on text classification, the collected evaluation data is divided into positive evaluation data, negative evaluation data, and neutral evaluation data, including:

[0018] Get the text that needs to be classified;

[0019] Use Word2Vec to convert the text to be classified into vectors;

[0020] Use vectors to train SVM classifiers;

[0021] Using the trained SVM classifier, the text to be classified is divided into positive evaluation data, negative evaluation data and neutral evaluation data.

[0022] Optionally, through predefined syntactic structures, identify demand evaluations and negative evaluations in the evaluation data, extract key information from the evaluation data, and generate a visual report, including:

[0023] Define the grammatical structure of demand and negative comments and extract key information;

[0024] The extracted key information is visualized using the Matplotlib tool.

[0025] Optionally, define the grammatical structure of demand and negative comments and extract key information, including:

[0026] Use sentiment analysis algorithms in NLP technology to calculate sentiment scores for comments;

[0027] Define a keyword list for positive and negative reviews, and use keyword matching to help differentiate them;

[0028] Use NLP tools to perform grammatical parsing and identify syntactic structures in comments;

[0029] Process demand-oriented review information and negative review information separately;

[0030] Convolutional neural network is used to extract features from evaluation data.

[0031] Optionally, use sentiment analysis algorithms in NLP technology to calculate sentiment scores for comments, including:

[0032] Calculate sentiment scores for reviews using VADER, TextBlob, or a deep learning model.

[0033] Optionally, use NLP tools to perform syntax parsing and identify the grammatical structure in the comments, including:

[0034] Use the spaCy tool to identify grammatical structures in the reviews.

[0035] According to two aspects of the present invention, there is provided an interface design device, the interface design device comprising:

[0036] The collection module is used to automatically collect evaluation data on interface design on the Internet using crawler tools;

[0037] A differentiation module is used to differentiate the collected evaluation data into positive evaluation data, negative evaluation data and neutral evaluation data based on text classification;

[0038] An extraction module, used to identify demand evaluations and negative evaluations in the evaluation data through predefined syntactic structures, extract key information in the evaluation data, and generate a visual report;

[0039] A calculation module is used to set the weight coefficient of each function and calculate the function priority based on the number and rating of user comments;

[0040] The verification module is used to verify the feasibility of the design process through A / B testing.

[0041] According to three aspects of the present invention, there is provided an electronic device, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus;

[0042] A computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the interface design method.

[0043] According to four aspects of the present invention, there is provided a computer-readable storage medium, comprising: a computer program executable by an electronic device is stored therein, and when the computer program runs on the electronic device, the electronic device executes the steps of the interface design method.

[0044] Through the above scheme, the following beneficial technical effects are obtained:

[0045] The present invention provides an efficient intelligent cockpit interface optimization method by combining automated data collection, natural language processing, information extraction and data visualization. The technical points to be protected include data collection and processing methods, text classification and semantic analysis technology, data visualization implementation and function priority calculation method. These technical points not only improve the efficiency of user demand analysis, but also provide a scientific basis for interface design and enhance the market competitiveness of products. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flow chart of an interface design method provided by one or more embodiments of the present invention.

[0047] Figure 2 It is a flow chart of the collection operation in the interface design method provided by one or more embodiments of the present invention.

[0048] Figure 3 It is a flowchart of crawling web page information in a collection operation provided by one or more embodiments of the present invention.

[0049] Figure 4 It is a flow chart for distinguishing operations in the interface design method provided by one or more embodiments of the present invention.

[0050] Figure 5 It is a flowchart of extracting operations in the interface design method provided by one or more embodiments of the present invention.

[0051] Figure 6 It is a flowchart of the grammatical structure definition in the extraction operation provided by one or more embodiments of the present invention.

[0052] Figure 7 It is a structural diagram of an interface design device provided by one or more embodiments of the present invention.

[0053] Figure 8 is a structural diagram of an electronic device provided by one or more embodiments of the present invention. DETAILED DESCRIPTION

[0054] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. 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.

[0055] Human-Machine Interaction (HMI) refers to the information exchange process between people and computers using a certain dialogue language and a certain interactive method to complete a certain task. Many well-known companies and academic institutions are studying human-computer interaction. In the history of computer development, people rarely paid attention to the usability of computers. Now, many computer users complain that computer manufacturers do not put enough effort into making their products "user-friendly". In turn, these computer system developers are also complaining. Their reason is that designing and manufacturing computers is a very complicated task. Just studying how to apply computers in new fields has occupied most of their energy. There is really no extra energy to study how to improve the usability of computers. An important issue of human-computer interaction (HMI) is that different computer users have different usage styles-they have different educational backgrounds, different ways of understanding, different learning methods, and different skills. For example, the usage habits of a left-handed person and an ordinary person are completely different. In addition, cultural and ethnic factors must also be considered. Secondly, the research and design of human-computer interaction needs to consider that the user interface technology changes rapidly, and the new interaction technology provided may not be applicable to previous research. Also, as users gradually master the new interface, they may make new requirements.

[0056] User Interface refers to the overall design of the software's human-computer interaction, operating logic, and aesthetic interface. A good UI design not only makes the software personalized and tasteful, but also makes the software's operation comfortable, simple, free, and fully reflects the software's positioning and characteristics. The user interface (abbreviated as UI, also known as the user interface) is the medium for interaction and information exchange between the system and the user. It realizes the conversion between the internal form of information and the form acceptable to humans. The user interface is software designed between the user and the hardware for interaction and communication between them. The purpose is to enable users to operate the hardware conveniently and efficiently to achieve two-way interaction and complete the work they hope to complete with the help of the hardware. The definition of the user interface is broad, including human-computer interaction and graphical user interface. User interfaces exist in all fields involving information exchange between humans and machines.

[0057] Figure 1 A flowchart of an interface design method provided by one or more embodiments of the present invention is shown. Figure 1, the interface design method includes the following steps:

[0058] S11, uses crawler tools to automatically collect evaluation data on interface design on the Internet.

[0059] S12, based on text classification, the collected evaluation data is divided into positive evaluation data, negative evaluation data and neutral evaluation data.

[0060] S13, through the predefined syntactic structure, identify the demand evaluation and negative evaluation in the evaluation data, extract the key information in the evaluation data, and generate a visual report.

[0061] S14, based on the number and ratings of user comments, weighting coefficients of various functions are set and function priorities are calculated.

[0062] S15, verify the feasibility of the design process through A / B testing.

[0063] To ensure the authenticity and strong guidance of user information, it is necessary to collect a large number of evaluations from car owners or car enthusiasts.

[0064] For the collection of data samples, if we rely solely on manpower to collect information, it will not only be inefficient and cumbersome, but also increase the cost of collection. We can use crawler tools (such as BeautifulSoup and Selenium), also known as network robots, to help automatically collect data information.

[0065] In order to obtain data in the automotive vertical field, a large number of professional automotive evaluation websites need to be read and collected. At the same time, in order to ensure data quality and availability, some useless information such as web page advertisements often need to be filtered out. We will propose web page features that can be extracted, such as "in-vehicle HMI", "user experience", etc., and use whether the proposed features are met as the basis for page classification to collect available web pages. At the same time, in order to ensure that the number of collected samples is large enough, the proposed features will be identified and calculated to query the number of samples.

[0066] There are also some irrelevant or ambiguous comments on professional automotive websites; data preprocessing can perform preliminary identification and error correction on the collected batch comments, including sentence segmentation and spelling correction.

[0067] For user comments collected by the crawler, perform sentence processing and delete noise words and non-Chinese characters.

[0068] After obtaining the pre-processed comments, it is necessary to conduct a preliminary classification of the user comments; distinguish between positive / negative / neutral evaluation categories.

[0069] In the process of processing a large number of reviews, the model needs to understand our text information; word vectors can map the user's vocabulary into vector form, so the processing of user comments can be converted from language semantic classification to calculation between vectors.

[0070] Therefore, word embedding and support vector machine (SVM) can be used to classify user comments and filter out constructive comments related to UI interface improvement.

[0071] Based on the above training, we have obtained classified user reviews. How to enhance the readability of the reviews? Through predefined syntactic structures, we can extract key information from the reviews and generate a visual report.

[0072] Define the grammatical structure of demand and negativity comments and perform feature extraction.

[0073] Grammatical structure of demanding and negative comments:

[0074] The grammatical structure of a demand review:

[0075] Demand comments usually include users' requests or suggestions for certain features or services. Its grammatical structure includes:

[0076] Verb phrases: Verbs that express a wish, suggestion, or request (e.g., "hope," "suggest," "need").

[0077] Noun phrase: This refers to a specific function or feature (e.g., "better navigation system," "greater safety").

[0078] Qualifiers: Attributes that describe the desired functionality or service (e.g., “more efficient,” “more flexible”).

[0079] Example:

[0080] “I wish the navigation system was more accurate.”

[0081] “It is recommended to add an autonomous driving function.”

[0082] The grammatical structure of a negative review:

[0083] Negative comments usually express users’ dissatisfaction or negative emotions. Their grammatical structure includes:

[0084] Subject: The user or subject of a specific function (such as "I", "system").

[0085] Verb phrases: Verbs that express negative emotions or problems (e.g., “hate,” “breakdown,” “slow”).

[0086] Noun phrase: A noun that describes a specific function or feature (e.g., "interface," "performance").

[0087] Emotion words: adjectives or adverbs that express negative emotions (e.g., "terrible," "very slow").

[0088] Example:

[0089] “I hate that the system keeps breaking down.”

[0090] “The interface is very slow and the user experience is poor.”

[0091] (2) Differences in grammatical structure between demand and negative comments

[0092] Verb Use:

[0093] The verbs used in demand-based comments are usually words that express requests, such as hope, suggestion, and need.

[0094] Verbs used in negative comments are usually words that express dissatisfaction or problems, such as hate, breakdown, slow, etc.

[0095] Emotional expression:

[0096] Demand-based comments usually contain constructive suggestions and are more neutral or positive in sentiment.

[0097] Negative comments usually carry strong negative emotions, and sentiment words are obviously negative.

[0098] Target audience:

[0099] Need-based reviews are usually targeted at features or services that are expected to be improved.

[0100] Negative reviews are usually targeted at features or functions that users consider problematic.

[0101] Provide low-code or zero-code designers with charts and interactive interfaces to display analysis results, helping them quickly understand market trends and user needs.

[0102] As the earliest data visualization library in Python, Matplotlib provides a wealth of drawing functions and flexible configuration options that can meet most basic drawing needs; it can visualize the extracted key information.

[0103] In terms of the basics of graph drawing, we need to master several core concepts, including coordinate axes, legends, titles, labels, etc. Coordinate axes are the basis of charts and are used to show the distribution of data; legends are used to distinguish different series of data; titles and labels are used to explain the theme and content of the chart.

[0104] Statistics on user complaints of each module of the car IVI; we use pie charts for this statistics; because pie charts are suitable for showing the proportion of each category and for relative comparison of data.

[0105] (1) Import library:

[0106] importmatplotlib.pyplot as plt

[0107] (2) Prepare data:

[0108] labels = ['A','B','C','D']

[0109] sizes = [15,30,45,10]

[0110] (3) Draw a pie chart:

[0111] plt.pie(sizes,labels=labels)

[0112] plt.show()

[0113] The above pie chart shows the percentage of complaints in each module.

[0114] In automotive HMI design, in order to ensure driving safety and optimize passenger entertainment experience, we need to reduce the interface level and enhance interface usability. Based on the star rating of user reviews and the frequency of function usage, we can calculate the optimization priority of each function and provide a theoretical basis for interface level design.

[0115] Based on the number and ratings of user reviews, the weighting coefficients of each function are set and the function priority is calculated.

[0116] The priority algorithm needs to consider different weight settings and scoring mechanisms when dealing with demand and negative reviews. The following will discuss in detail the differences in weight settings, the basis for weight coefficient design, and the calculation method of scoring.

[0117] (1) Differences in weight settings

[0118] The weight settings for demand reviews and negative reviews are different as follows:

[0119] Demand review weight:

[0120] The weights are mainly based on the urgency of the user's need and the feasibility of the suggestion.

[0121] For example, a requirement that occurs frequently may be weighted more highly than a requirement that occurs less frequently.

[0122] Negative review weight:

[0123] The weighting is based primarily on the severity of the issue and the degree of negative impact on the user experience.

[0124] For example, a problem that causes the system to crash will be weighted higher than a problem that only affects the aesthetics of the interface.

[0125] (2) Design basis of weight coefficient

[0126] The design of weight coefficients can be based on the following aspects:

[0127] Frequency: The number of times a comment appears or the proportion of total comments.

[0128] User impact: The degree to which the issues or needs mentioned in the review affect the user experience, which can be measured through user feedback or survey results.

[0129] Solution Difficulty: The complexity and resource requirements of the need or problem to be solved.

[0130] Urgency: How quickly a need or problem needs to be addressed.

[0131] (3) Calculation method of scoring

[0132] The score can be calculated using a weighted scoring model, combining the various weight coefficients to get the total score. The formula is as follows:

[0133] Rating = ∑(weight coefficient × related index)

[0134] Specifically, you can use the following steps:

[0135] 1. Determine the weight coefficient:

[0136] Frequency weight: Weight frequency

[0137] User influence weight: Weight impact

[0138] Solving difficulty weight: Weight difficulty

[0139] Urgency Weight: Weight urgency

[0140] 2. Calculate relevant indicators:

[0141] Frequency

[0142] User Impact: Impact

[0143] Difficulty

[0144] Urgency: Urgency

[0145] 3. Calculate the score:

[0146] Rating = (Weight frequency ×frequency)+(Weight impact ×impact)+(Weight difficulty ×difficulty)

[0147] +(Weight urgency ×urgency)

[0148] Priority calculation instructions:

[0149] defcalculate_priority(comments):

[0150] weight_factors = [1.5, 1.25, 1, 1.25, 1.5] # weight coefficient

[0151] priorities={}

[0152] forfeature,ratings incomments.items():

[0153] priority=sum(r*weight_factors[i]fori,rinenumerate(ratings))

[0154] priorities[feature]=priority

[0155] returnpriorities

[0156] The above can be used to derive the weight coefficient of each function, which can be used as a reference for the design of the human-computer interaction interface level.

[0157] The designer compares the design after adopting the above scheme with the original design to confirm the feasibility of the scheme.

[0158] In order to ensure the number of user samples and not affect the HMI design that is currently in mass production, we use A / B testing to conduct pre-launch testing.

[0159] In the same time dimension, car owners / car enthusiasts are randomly selected to form the same (similar) visitor groups (target population) to randomly visit the A / B versions, and the user experience data and business data of each group are collected. The current design plan is officially adopted after being tested.

[0160] 1. Determine the test objectives: It is necessary to determine the test objectives, such as improving the accuracy of interface operations, reducing the operation levels, or improving user satisfaction.

[0161] 2. Select test variables: Select the variables to be tested, such as the multimedia module of the HMI interface, navigation color scheme, etc.

[0162] 3. Develop a test plan: Develop a test plan, including test time, number of test samples, test plan, etc.

[0163] 4. Create a test version: Select two stable software versions before and after the solution is applied.

[0164] 5. Randomly assign users: Randomly assign website users into two or more groups, and each group is shown a different version of the test variable.

[0165] 6. Collect data: To collect user behavior and feedback data, we can use eye trackers and electroencephalography; capture objective data such as user gaze rate; we can also use user interviews and other methods to understand users' subjective evaluations.

[0166] 7. Analyze data: Analyze the data, compare the performance of each group, and determine whether the version of the application solution is optimized.

[0167] 8. Apply results: Based on the test results, apply the most effective version to optimize HMI design and content.

[0168] Continuous improvement: A / B testing is an ongoing process; user behavior and feedback data can be regularly monitored and analyzed to optimize and improve different HMI modules and improve the effectiveness of products or services.

[0169] In automotive HMI design, through automated data collection, preprocessing, classification analysis and visualization methods, the interface design guided by the designer's subjective aesthetic can be optimized into a scientific interface design based on data and user-oriented. This can not only simplify the interface designer's workflow and shorten the R&D cycle, but also meet the needs of users and enhance user experience and market competitiveness.

[0170] Figure 2 A flowchart of the collection operation in the interface design method provided by one or more embodiments of the present invention is shown. Figure 2 , using crawler tools to automatically collect evaluation data on interface design on the Internet, including the following steps:

[0171] S21, uses crawler tools to automatically collect evaluation data.

[0172] S22, perform preliminary identification and error correction on the collected batch comments.

[0173] The process of automatically collecting comment information on web pages through crawler tools generally includes the following steps:

[0174] (1) Read / access web pages

[0175] Import BeautifulSoup and a parser; import the requests library to read web page content from the network:

[0176] frombs4 importBeautifulSoup

[0177] importrequests

[0178] from selenium import webdriver

[0179] url = 'https: / / example.com'

[0180] response = requests.get(url)

[0181] (2) Collection website

[0182] Use BeautifulSoup and a parser to scrape website content:

[0183] soup=BeautifulSoup(response.text,'html.parser')

[0184] (3) Web page feature extraction

[0185] Use selectors to extract data about defined web page features; for example, use the .find() or .find_all() method to find the tag "Car HMI" and use the .get_text() method to get the text within the tag:

[0186] divs_with_class=soup.find_all('div',class_='some-class')

[0187] (4) Identification feature calculation parameters

[0188] Use the count function to count the feature data extracted by the selector:

[0189] def popular_words(text, words):

[0190] lower_count=text.lower().split().count

[0191] return{word:lower_count(word)forword inwords}

[0192] The above preprocessing process is as follows:

[0193] (1) Participle

[0194] Word segmentation is sentence segmentation. In order to accurately obtain user semantics, reasonable sentence segmentation is required. For example, "I like the car voice assistant" can be divided into I / like / car / voice / assistant / ; I like / car voice assistant / , etc. When segmenting words, we should try to segment words with large granularity. That is, the more words a word has, the more accurate the meaning it can express. In the case of the same number of words, the fewer the total number of words, the fewer semantic units, then the relative weight of a single semantic unit will be greater, and therefore the accuracy will be higher.

[0195] Use jieba.Tokenizer(dictionary = DEFAULT_DICT) to create a custom tokenizer, which can be used to use different dictionaries at the same time. jieba.dt is the default tokenizer, and all global tokenizer-related functions are mappings to this tokenizer.

[0196] seg_list = jieba.cut("I like the car voice system", cut_all = True)

[0197] print("Full Mode:"+" / ".join(seg_list))

[0198] (2) Cleaning

[0199] Data cleaning can help us remove useless data and reduce the size of training data, including removing punctuation, stop words, rare words, etc. Because the above does not add any additional information to the text data, there is no need to retain it.

[0200] importre

[0201] defpreprocess_text(text):

[0202] text=re.sub(r'[^\u4e00-\u9fa5]',",text)#Only keep Chinese characters

[0203] sentences = text.split('.')

[0204] return sentences

[0205] Remove stop words / scarce words:

[0206] A list stopwords can be created as one's own stopword library or a predefined library can be used.

[0207] train['tweet'] = train['tweet'].apply(lambda sen: "".join(x for x in sen.split() if x not in stop))

[0208] (3) Standardization

[0209] Standardization means spelling correction. The main task is to map incorrect words to correct words. For example, "removing the machine" is corrected to "car machine". At the same time, the incorrect words are compared with the words in the dictionary to find the most matching word.

[0210] Characters are deleted one by one from the incorrect word to generate all possible subsequences.

[0211] These subsequences are compared with the words in the dictionary to find the most matching word.

[0212] from textblob import TextBlob

[0213] train['tweet'][:5].apply(lambda x: str(TextBlob(x).correct()))

[0214] Figure 3 The flowchart of web page information crawling in the collection operation provided by one or more embodiments of the present invention is shown. Refer to Figure 3 , a crawler tool is adopted to help automatically collect data information, including the following operation steps:

[0215] S31, read / access the web page.

[0216] S32, collect comment data from the website.

[0217] S33, extract the web page collection features.

[0218] S34, store the extracted web page collection features into the web page database.

[0219] S35, obtain the page features.

[0220] S36, perform page classification according to the extracted web page collection features and the page features.

[0221] S37, specify the page features.

[0222] S38, identifying characteristic calculation parameters.

[0223] S39, count the number of queries.

[0224] In various embodiments of the present application, the process of automatically collecting comment data on a web page using a crawler tool mainly consists of the following operating steps.

[0225] First, the target web page containing the interface comment data is read and accessed.

[0226] After completing the reading / accessing of the target web page, the collection of comment data is performed on the target website according to the read information.

[0227] After the review data is collected, the web page collection features are extracted.

[0228] It should be understood that different web pages have different characteristics for the collection of comment data. For example, different websites have their corresponding types. Some websites are car sales websites, from which real-time transaction data of vehicles can be easily captured. Other websites are car review websites, and many car enthusiasts have published a large number of user experiences of specific models on these review websites. On car review websites, it is easy to obtain the public's preference data for different models.

[0229] Not only the type of website, but also some other features closely related to data collection operations can be classified as the category of stolen web page collection features. For example, for web pages implemented with JSP technology, the response speed is generally faster than that of web pages implemented with PHP and ASP technology. Therefore, when collecting comment data on web pages implemented with JSP, the waiting time for collection can be set relatively short. For data collection of web pages implemented with PHP and ASP technology, the setting value of the waiting time should be relatively large.

[0230] After the web page collection features are extracted, the page is classified according to the read web page data. The goal of the above classification is to divide the comment data in the web page into: positive comment data, negative comment data, and neutral comment data according to the sentiment tendency of the text in the web page.

[0231] The above-mentioned classification operation of pages utilizes the page features of the classified pages.

[0232] It should be understood that in the process of collecting comment data, some page features of the target web page can be obtained. These page features may include: the specific location of the target comment area on the web page, the location of the dynamic comment area, and so on.

[0233] These page features can be obtained through the experience summary of multiple data collection actions, that is, the statistics of historical collection data. For example, for a target web page, data collection is performed in different time periods, and it is found that the text contained in some specific areas of the web page is dynamically changing. Based on the dynamic changes of such historical data, it can be determined that the area containing dynamically changing text data is the dynamic comment area in the web page, which is the main target area of ​​the comment data collection action.

[0234] The acquisition of dynamic comment areas similar to the above can be considered as an action of extracting page features.

[0235] After completing the classification of positive comment data, negative comment data, and neutral comment data, you can specify page features and count the feature data extracted by the selector based on the specified page features, that is, identify feature calculation parameters.

[0236] Furthermore, after the identification of the characteristic calculation parameters is completed, the number of web page queries can be further counted based on the identified characteristic calculation parameters.

[0237] Figure 4 FIG. 1 is a flowchart showing the distinguishing operation in the interface design method provided by one or more embodiments of the present invention. Figure 4 , based on text classification, the collected evaluation data is divided into positive evaluation data, negative evaluation data and neutral evaluation data, including the following steps:

[0238] S41, obtaining text to be classified.

[0239] S42, using Word2Vec, converts the text to be classified into a vector.

[0240] S43, train the SVM classifier using the vector.

[0241] S44, using the trained SVM classifier, the text to be classified is divided into positive evaluation data, negative evaluation data and neutral evaluation data.

[0242] In by Figure 4 In the illustrated embodiment, the distinction between positive evaluation data, negative evaluation data, and neutral evaluation data is accomplished by a pre-built SVM classifier.

[0243] Support Vector Machine (SVM) is a generalized linear classifier that performs binary classification of data in a supervised learning manner. Its decision boundary is the maximum-margin hyperplane solved for the learning samples.

[0244] SVM uses the hinge loss function to calculate the empirical risk and adds a regularization term to the solution system to optimize the structural risk. It is a classifier with sparsity and robustness. SVM can perform nonlinear classification through the kernel method and is one of the common kernel learning methods.

[0245] The target of SVM classification is the comment data captured from the website. In other words, the object of SVM classification is the text obtained from the Internet.

[0246] Before classifying the actual classification target, that is, text, it is necessary to use the Word2Vec tool to convert the text into a vector.

[0247] After the above-mentioned vector conversion is completed, the obtained vector is labeled with corresponding properties. That is, according to the original attributes of the text data, the converted vector is labeled as positive evaluation data, negative evaluation data and neutral evaluation data.

[0248] After the training data is labeled, the labeled data is input into the SVM classifier to complete the SVM classification training.

[0249] After the above-mentioned classifier training and parameter tuning, an SVM classifier that can actually classify the original data has been obtained. At this point, the trained SVM classifier is applied to the classification of the original comment data to obtain the classification result of whether the original comment data belongs to positive comment data, negative comment data, or neutral comment data.

[0250] Figure 5 FIG. 1 is a flowchart showing an extraction operation in an interface design method provided by one or more embodiments of the present invention. Figure 5 , through the predefined syntactic structure, identify the demand evaluation and negative evaluation in the evaluation data, extract the key information in the evaluation data, and generate a visual report, including:

[0251] S51, define the grammatical structure of demand and negative comments, and extract key information.

[0252] S52, using the Matplotlib tool, the extracted key information is visualized.

[0253] In various embodiments of the present invention, the key information extraction operation mainly includes two sub-operations, which are: feature extraction through grammatical structure definition, and visualization operation of the extracted key information.

[0254] As for the definition of grammatical structure, we define the unique grammatical structure of demand comments and negative comments respectively according to their characteristics.

[0255] For example, demand reviews often contain verbs that express hope, suggestion, or request. For example, such verbs may include: hope, suggestion, need. Therefore, in the process of identifying demand reviews, it is possible to prioritize whether the review contains some of the above verbs. If verbs with the above characteristics appear, the review can be determined to be a demand review.

[0256] For another example, negative comments often contain some verbs that express negative emotions or problems. These verbs can be: hate, collapse, slow. Therefore, in the process of identifying negative comments, these verbs that express negative emotions or problems can be identified. If the above verbs are identified, the above comment can be determined to be a negative comment.

[0257] After identifying demand reviews and negative reviews, key information can be extracted from the text data of the above two types of reviews respectively.

[0258] In various embodiments of the present application, the tool used to extract the above key information is a convolutional neural network. A convolutional neural network is a deep learning model for image and text data, which is mainly composed of a convolutional layer, a pooling layer, and a fully connected layer. The convolutional layer is used to extract features from text data, the pooling layer is used to reduce dimensionality and reduce the amount of calculation, and the fully connected layer is used for classification prediction. After we design the key information definition, the user evaluation related feature extraction is mainly completed through the convolutional layer.

[0259] After completing the extraction of the above key information, the above key information can be visualized and various visualization reports can be generated to make the extracted features more intuitive for users.

[0260] Figure 6 FIG. 1 is a flowchart showing the definition of a grammatical structure in an extraction operation provided by one or more embodiments of the present invention. Figure 6, define the grammatical structure of demand and negative comments, and perform feature extraction, including the following steps:

[0261] S61, using the sentiment analysis algorithm in NLP technology to calculate the sentiment score of the comments.

[0262] S62, define a keyword list of demand and negative comments, and assist in distinguishing them through keyword matching.

[0263] S63, use NLP tools to perform grammatical parsing and identify the syntactic structure in the comments.

[0264] S64, processing the demand review information and the negative review information separately.

[0265] S65, uses convolutional neural network to extract features from evaluation data.

[0266] (1) Grammatical structure of demand and negative comments

[0267] The grammatical structure of a demand review:

[0268] Demand comments usually include users' requests or suggestions for certain features or services. Its grammatical structure includes:

[0269] Verb phrases: Verbs that express a wish, suggestion, or request (e.g., "hope," "suggest," "need").

[0270] Noun phrase: This refers to a specific function or feature (e.g., "better navigation system," "greater safety").

[0271] Qualifiers: Attributes that describe the desired functionality or service (e.g., “more efficient,” “more flexible”).

[0272] Example:

[0273] “I wish the navigation system was more accurate.”

[0274] “It is recommended to add an autonomous driving function.”

[0275] The grammatical structure of a negative review:

[0276] Negative comments usually express users’ dissatisfaction or negative emotions. Their grammatical structure includes:

[0277] Subject: The user or subject of a specific function (such as "I", "system").

[0278] Verb phrases: Verbs that express negative emotions or problems (e.g., “hate,” “breakdown,” “slow”).

[0279] Noun phrase: A noun that describes a specific function or feature (e.g., "interface," "performance").

[0280] Emotion words: adjectives or adverbs that express negative emotions (e.g., "terrible," "very slow").

[0281] Example:

[0282] “I hate that the system keeps breaking down.”

[0283] “The interface is very slow and the user experience is poor.”

[0284] (2) Differences in grammatical structure between demand and negative comments

[0285] Verb Use:

[0286] The verbs used in demand-based comments are usually words that express requests, such as hope, suggestion, and need.

[0287] Verbs used in negative comments are usually words that express dissatisfaction or problems, such as hate, breakdown, slow, etc.

[0288] Emotional expression:

[0289] Demand-based comments usually contain constructive suggestions and are more neutral or positive in sentiment.

[0290] Negative comments usually carry strong negative emotions, and sentiment words are obviously negative.

[0291] Target audience:

[0292] Need-based reviews are usually targeted at features or services that are expected to be improved.

[0293] Negative reviews are usually targeted at features or functions that users consider problematic.

[0294] (3) Methods for distinguishing between positive and negative comments

[0295] 1. Sentiment Analysis

[0296] Use sentiment analysis algorithms from NLP technologies (such as VADER, TextBlob, or deep learning models) to calculate sentiment scores for the reviews. Based on the scores, distinguish between positive, neutral, and negative reviews.

[0297] 2. Keyword detection

[0298] Define a keyword list for positive and negative reviews, and use keyword matching to assist in differentiation.

[0299] 3. Syntax analysis

[0300] Use NLP tools (such as spaCy) to perform grammatical analysis and identify the syntactic structure in the comments. Determine the comment type based on the grammatical structure.

[0301] Sentiment analysis can be used as the most commonly used preliminary differentiation method to roughly divide comments into three categories: positive, neutral, and negative. For comments with score ranges close to positive and neutral or neutral and negative, we can choose custom keywords for secondary differentiation to improve the classification granularity. Syntax parsing can re-examine and judge the comments differentiated by the above two methods to improve classification accuracy.

[0302] 4. How to handle demand and negative review information

[0303] Demand comment processing:

[0304] Priority Assessment:

[0305] Assess the priority of features or functions based on their importance and frequency of mention in reviews.

[0306] Improvement plan formulation:

[0307] Develop corresponding product improvement plans for high-priority demand reviews.

[0308] User Feedback Loop:

[0309] Feedback improvement plans and progress to users and establish good user communication channels.

[0310] Dealing with negative comments:

[0311] Problem location:

[0312] Analyze the issues mentioned in the negative reviews to determine their specific manifestations and frequency.

[0313] Root Cause Analysis:

[0314] Through further data analysis and user investigation, the root cause of the problem was found.

[0315] Solution:

[0316] Develop solutions to the root causes and optimize products or services.

[0317] Improved user satisfaction:

[0318] After solving the problem, actively provide feedback to users to improve user satisfaction.

[0319] In addition, convolutional neural networks were further used to extract features from the evaluation data.

[0320] Convolutional Neural Network (CNN) is a deep learning model that has achieved great success in the field of computer vision. Their design is inspired by the visual system in biology and aims to simulate the way human vision is processed. In the past few years, CNN has made significant progress in image recognition, object detection, image generation and many other fields, becoming an important part of computer vision and deep learning research.

[0321] The convolution kernel is a weight matrix that generates a feature map by sliding summation with the input text data; thereby extracting the specified features we need.

[0322] Mathematical model formula:

[0323]

[0324] Among them, y ij is the value of the i-th row and j-th column of the output feature map, x ik is the value of the i-th row and k-th column of the input text data, w kj is the value of the kth row and jth column of the convolution kernel, b j is the bias term, and K is the width of the convolution kernel.

[0325] import spacy

[0326] nlp=spacy.load('zh_core_web_sm')

[0327] defextract_information(sentence):

[0328] doc=nlp(sentence)

[0329] fortoken in doc:

[0330] iftoken.dep_ == 'ROOT':

[0331] print(f'Subject:{token.text},Predicate:{token.head.text}')

[0332] The above can extract the specified features.

[0333] like Figure 7 As shown, the present invention also provides an interface design device, comprising:

[0334] The collection module 71 is used to automatically collect evaluation data on interface design on the Internet using crawler tools.

[0335] The distinguishing module 72 is used to distinguish the collected evaluation data into positive evaluation data, negative evaluation data and neutral evaluation data based on text classification.

[0336] The extraction module 73 is used to identify demand evaluations and negative evaluations in the evaluation data through predefined syntactic structures, extract key information in the evaluation data, and generate a visual report.

[0337] The calculation module 74 is used to set the weighting coefficient of each function and calculate the function priority based on the number and rating of user comments.

[0338] The verification module 75 is used to verify the feasibility of the design process through A / B testing.

[0339] It is worth noting that although only some basic functional modules are disclosed in the embodiments of the present invention, it does not mean that the composition of the present system is limited to the above basic functional modules. On the contrary, what the present embodiment wants to express is that on the basis of the above basic functional modules, those skilled in the art can arbitrarily add one or more functional modules in combination with the prior art to form an infinite number of embodiments or technical solutions. In other words, the present system is open rather than closed. Just because the present embodiment only discloses individual basic functional modules, it cannot be considered that the protection scope of the claims of the present invention is limited to the disclosed basic functional modules. At the same time, for the convenience of description, the above devices are described in terms of functions and are described separately in various units and modules. Of course, when implementing the present invention, the functions of each unit and module can be implemented in the same or one or more software and / or hardware.

[0340] like Figure 8 As shown, the present invention also provides an electronic device, comprising: a processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the interface design method.

[0341] Figure 8 Schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. Figure 8 The electronic device provided in the embodiment of the present invention includes: one or more processors 810 and a memory 820; the processor 810 in the electronic device may be one or more, Figure 8 A processor 810 is taken as an example; the memory 820 is used to store one or more programs; the one or more programs are executed by the one or more processors 810, so that the one or more processors 810 implement the interface design method as described in any one of the embodiments of the present invention.

[0342] The electronic device may further include: an input device 830 and an output device 840 .

[0343] The processor 810, memory 820, input device 830 and output device 840 in the electronic device can be connected via a bus or other means. Figure 8 The example of connecting through bus is taken in the following.

[0344] The memory 820 in the electronic device is a computer-readable storage medium that can be used to store one or more programs, which can be software programs, computer executable programs, and modules, such as the program instructions / modules corresponding to the full-link test method for voice vehicle control provided in the embodiment of the present invention. The processor 810 executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory 820, that is, implements the interface design method in the above method embodiment.

[0345] The memory 820 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the electronic device, etc. In addition, the memory 820 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 820 may further include a memory remotely arranged relative to the processor 810, and these remote memories may be connected to the device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0346] The input device 830 may be used to receive input digital or character information and generate key signal input related to user settings and function control of the electronic device. The output device 840 may include a display device such as a display screen.

[0347] The present invention also provides a computer-readable storage medium storing a computer program executable by an electronic device, and when the computer program runs on the electronic device, the electronic device executes the steps of the interface design method.

[0348] Specifically, the computer storage medium of the embodiment of the present invention can adopt any combination of one or more computer-readable media. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples (non-exhaustive list) of computer-readable storage media include: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program, which can be used by an instruction execution system, device or device or used in combination with it.

[0349] The present invention also provides a vehicle, wherein the vehicle is provided with the interface design device as described above.

[0350] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An interface design method, characterized in that: The interface design method comprises: Use crawler tools to automatically collect evaluation data on interface design on the Internet; Based on text classification, the collected evaluation data is divided into positive evaluation data, negative evaluation data and neutral evaluation data; Through the predefined syntactic structure, identify the demand evaluation and negative evaluation in the evaluation data, extract the key information in the evaluation data, and generate a visual report; Based on the number and ratings of user reviews, set the weighting coefficients for each function and calculate the function priority; The feasibility of the design process was verified through A / B testing.

2. The method according to claim 1, characterized in that Use crawler tools to automatically collect evaluation data on interface design on the Internet, including: Use crawler tools to automatically collect evaluation data; Conduct preliminary identification and error correction on the collected batch reviews; Among them, crawler tools include: BeautifulSoup, or Selenium.

3. The method according to claim 1, characterized in that Based on text classification, the collected evaluation data is divided into positive evaluation data, negative evaluation data and neutral evaluation data, including: Get the text that needs to be classified; Use Word2Vec to convert the text to be classified into vectors; Use vectors to train SVM classifiers; Using the trained SVM classifier, the text to be classified is divided into positive evaluation data, negative evaluation data and neutral evaluation data.

4. The method according to claim 1, characterized in that Through the predefined syntactic structure, it can identify the demand evaluation and negative evaluation in the evaluation data, extract the key information in the evaluation data, and generate a visual report, including: Define the grammatical structure of demand and negative comments and extract key information; The extracted key information is visualized using the Matplotlib tool.

5. The method according to claim 4, characterized in that Define the grammatical structure of demand and negative reviews and extract key information, including: Use sentiment analysis algorithms in NLP technology to calculate sentiment scores for comments; Define a keyword list for positive and negative reviews, and use keyword matching to help differentiate them; Use NLP tools to perform grammatical parsing and identify syntactic structures in comments; Process demand-oriented review information and negative review information separately; Convolutional neural network is used to extract features from evaluation data.

6. The method according to claim 5, characterized in that Use the sentiment analysis algorithm in NLP technology to calculate the sentiment score of the comments, including: Calculate sentiment scores for reviews using VADER, TextBlob, or a deep learning model.

7. The method according to claim 5, characterized in that Use NLP tools to perform grammatical parsing and identify grammatical structures in comments, including: Use the spaCy tool to identify grammatical structures in the reviews.

8. An interface design device, characterized in that: The interface design device comprises: The collection module is used to automatically collect evaluation data on interface design on the Internet using crawler tools; A differentiation module is used to differentiate the collected evaluation data into positive evaluation data, negative evaluation data and neutral evaluation data based on text classification; An extraction module, used to identify demand evaluations and negative evaluations in the evaluation data through predefined syntactic structures, extract key information in the evaluation data, and generate a visual report; A calculation module is used to set the weight coefficient of each function and calculate the function priority based on the number and rating of user comments; The verification module is used to verify the feasibility of the design process through A / B testing.

9. An electronic device, characterized in that: include: A processor, a communication interface, a memory and a communication bus, wherein the processor, the communication interface and the memory communicate with each other via the communication bus; a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of any one of the methods described in claims 1 to 7.

10. A computer-readable storage medium, characterized in that: It stores a computer program executable by an electronic device, and when the computer program runs on the electronic device, the electronic device executes the steps of the method described in any one of claims 1 to 7.

Images