CN102955810B - A kind of Web page classification method and equipment - Google Patents

Abstract

The invention discloses a method and equipment for classifying webpages. The method utilizes the records in the existing URL category database to establish virtual hierarchical URLs and predict the categories of the hierarchical URLs. When the webpage to be classified needs to be classified, the URL category library is queried according to the URL of the webpage to be classified; if no matching URL is found, the URL category library is queried according to the upper URL of the URL, and when a matching URL is The predicted category of the queried URL determines the category of the webpage to be classified. In the present invention, the efficiency and success rate of web page classification are improved.

Description

Method and device for classifying web pages

technical field

The invention relates to the technical field of the Internet, in particular to a web page classification method and equipment.

Background technique

With the rapid development of mobile Internet technology, the number of mobile Internet users is increasing. Therefore, the behavior analysis of mobile Internet users has gradually become a research hotspot.

In the prior art, user behavior is usually analyzed according to access logs of mobile Internet users. Specifically, the access log of the mobile Internet user is stored in the WAP (Wireless Application Protocol, Wireless Application Protocol) gateway, and the URL (UniversalResourceLocator, Uniform Resource Locator) of the webpage visited by the user is recorded in the access log. The category of web pages visited by the user can be known, and then the behavior preference of the corresponding user can be learned.

Wherein, the existing web page classification method may include the following steps:

1. The crawler crawls the content of the webpage;

2. Analyze the content of the webpage and obtain the corresponding text;

3. Analyze the text and obtain keywords;

4. Use an algorithm model, such as a text classification algorithm model such as Naive Bayesian or SVM, to classify; wherein, the algorithm model is usually trained in advance based on the training set.

The webpages (or the URLs corresponding to the webpages) visited by the user can be classified through the above method, and then a URL category library can be established. Wherein, the URL category library in the prior art may be as shown in Table 1.

Table 1

In the process of realizing the present invention, the inventor finds that there are at least the following problems in the prior art:

In the prior art, the URL category library is a simple flat data table without any relationship between entries. In order to accurately query the categories of webpages visited by users, a large amount of data needs to be stored and the category library needs to be updated in real time. And due to the rapid development of the Internet, the speed of adding web pages is extremely fast, even if the URL category library is updated once a day, the URL category library cannot save the classification of all web pages. At this time, the method that can be used is real-time crawling and prediction. It may take about tens of minutes to predict the category of a webpage. If batch prediction can be parallelized, the time is still very long, at least hours.

Contents of the invention

Embodiments of the present invention provide a method and device for classifying webpages, so as to improve the efficiency and success rate of determining webpage categories.

In order to achieve the above purpose, an embodiment of the present invention provides a method for classifying webpages, which is applied to the webpage classification process based on the URL category library. The URL category library records the URLs of each level and the predicted category of each URL. Among them, the adjacent The upper-level URL in the hierarchical URL is intercepted on the basis of the lower-level URL, and the method includes:

Query the URL category library according to the URL of the webpage to be classified;

If no matching URL is queried, the URL category library is queried according to the upper URL of the URL, and when a matching URL is queried, the category of the webpage to be classified is determined according to the predicted category of the queried URL.

The embodiment of the present invention also provides a webpage classification device, which is applied to the webpage classification process based on the Uniform Resource Locator URL category library. The URL category library records the URLs of each level and the predicted category of each URL. The upper-level URL in the hierarchical URL is intercepted on the basis of the lower-level URL, and the equipment includes:

The upper-level URL generation module is used to generate the upper-level URL of the URL according to the URL of the webpage to be classified;

The query module is used to query the URL category library according to the URL of the webpage to be classified; if no matching URL is found, then query the URL category library according to the upper URL of the URL;

A determining module, configured to determine the category of the webpage to be classified according to the predicted category of the queried URL when the query module inquires about a matching URL.

Compared with the prior art, the embodiment of the present invention divides the URLs into layers, records the URLs of each level in the URL category library, and records the predicted categories of each URL correspondingly; when it is necessary to determine the category of the webpage to be classified, obtain the Classify the URL of the web page, and check whether the URL is recorded in the URL category library; when the same URL is not recorded in the URL category library, it is determined as the category of the webpage to be classified according to the predicted category of the upper URL of the URL, which improves the determination Efficiency and success rate of web categories.

Description of drawings

FIG. 1 is a schematic diagram of a flow chart for generating a URL category library provided by an embodiment of the present invention;

FIG. 2 is a schematic flow chart of a method for classifying webpages provided by an embodiment of the present invention;

FIG. 3 is a schematic structural diagram of a webpage classification device provided by an embodiment of the present invention.

Detailed ways

Aiming at the defects in the prior art, the embodiment of the present invention proposes a technical solution for classifying webpages. In the technical solution proposed by the embodiment of the present invention, the URL is divided into levels by intercepting the URL, and the upper-level URLs of adjacent levels of URLs are obtained by intercepting the lower-level URLs, and the upper-level URLs are added to the existing URL category library. The record of URL (that is, URL, the prediction category of this URL and the upper URL of the adjacent hierarchy of this URL are recorded in the URL category storehouse in the embodiment of the present invention), and the prediction category of the record upper URL, when needing to classify the webpage, The URL category library can be queried according to the URL of the webpage to be classified; if no matching URL is queried, the URL category library is queried according to the upper URL of the URL, and when a matching URL is queried, it is determined according to the predicted category of the queried URL The category of the webpage to be classified, that is, when the URL of the webpage to be classified is not recorded in the URL category library, the category of the webpage to be classified can be determined according to the predicted category of the upper URL of the URL, and the URL corresponding to the upper URL of the URL to be classified can be checked. record, and use the predicted category of its upper-level URL as the predicted category of the webpage to be classified, which improves the efficiency and success rate of determining the category of the webpage.

Among them, the hierarchical division of the URL by intercepting the URL may be specifically implemented in the following manner:

According to the delimiter "/" in the URL, the URL is hierarchically divided, and the "/" is obtained sequentially from the end of the URL, and the preset number (such as 1) of the URL from the end to the field before the "/" It is the upper-level URL (ie, the upper-level URL) of the adjacent level of the URL.

For example, for the URL: http://3g.sina.com.cn:80/3g/static/sina.gif? t1=1252192802, http://3g.sina.com.cn:80/3g/static/sina.gif? t1=1252192802 is the first level of the URL, http://3g.sina.com.cn:80/3g/static/ is the second level of the URL, http://3g.sina.com.cn:80 /3g/ is the third level of the URL, http://3g.sina.com.cn: 80/3g/static/ is the upper level URL of the original URL, http://3g.sina.com.cn :80/3g/ is the upper level URL of http://3g.sina.com.cn:80/3g/static/.

It should be recognized that the manner of determining the upper-level URL in the technical solution proposed by the embodiment of the present invention is not limited to the foregoing manner, and may also be other manners.

The technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Apparently, the described embodiments are part of the embodiments of this application, not all of them. Based on the embodiments in the present application, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present application.

As shown in Figure 1, it is a schematic diagram of the establishment process of the URL category library proposed by the embodiment of the present invention. For the convenience of description, the URL information stored in the form of a data table in the URL category is used as an example for description, and each URL corresponds to an entry , the process of establishing the URL category library may include the following steps:

Step 101. Record the entry corresponding to the lowest-level URL in the URL category library. Wherein, the entry corresponding to the URL records the URL, the predicted category of the URL, and the upper-level URL of the URL.

Specifically, the URL of the webpage visited by the user in the past period of time (such as one month) can be used as the lowest-level URL in the URL category library, and the predicted category of the corresponding URL can be obtained through the existing webpage classification method; or, The URLs corresponding to some well-known websites are used as seeds, and a certain number of URLs are obtained through crawler crawling, and the obtained URLs are used as the lowest-level URLs in the URL category library, and the corresponding URLs are obtained through the existing webpage classification method. predicted category. After obtaining the lowest-level URL and its predicted category in the URL category library, obtain the upper-level URL of each lowest-level URL, and record the corresponding information (URL prediction category, upper-level URL) corresponding to the URL into the URL category library.

Step 102. Select an entry from the URL category library, and obtain the upper-level URL of the URL recorded in the entry.

Specifically, the table items in the URL category library are traversed, and the table items in the URL category library are sequentially selected to obtain the upper-level URL in the selected table item.

Step 103, judging whether an entry corresponding to the upper-level URL is stored in the URL category library. If the judgment is yes, go to step 102; otherwise, go to step 104.

Specifically, when the table item corresponding to the upper-level URL is stored in the URL category library, another table item is selected again; when the table item corresponding to the upper-level URL is not stored in the URL category library, you need Create an entry corresponding to the upper-level URL.

Step 104: Determine the predicted category of the upper-level URL and the upper-level URL of the upper-level URL, and record them in the URL category library.

Specifically, the entries in the URL category library are traversed to obtain entries with the same upper-level URL, and the predicted category of the upper-level URL is determined according to the predicted category of URLs in the obtained entries.

Wherein, determining the predicted category of the upper-level URL may specifically be implemented in the following manner:

Obtain all URLs whose upper-level URLs are the URLs of the category to be predicted from the URL category library; determine the quantity of URLs of each prediction category in the obtained URLs; determine the prediction category with the largest number of URLs as the category to be predicted The predicted category of the URL for the predicted category.

For example, for the following 4 URLs:

http://www.chinaweekly.cn/bencandy.php? fid=48&id=5464 Prediction category: history

http://www.chinaweekly.cn/bencandy.php? fid=48&id=5463 Prediction category: history

http://www.chinaweekly.cn/bencandy.php? fid=48&id=5344 Prediction category: history

http://www.chinaweekly.cn/bencandy.php? fid=49&id=5449 Forecast category: current comment

These four URLs have the same upper-level URL: http://www.chinaweekly.cn/. Since the upper-level URL is adjacent to the lower-level URL, there are 3 prediction categories as history and 1 prediction category as current commentary. Therefore, the predicted category of the upper-level URL is history.

It should be noted that, in the technical solution provided by the embodiment of the present invention, the predicted probability of each URL may also be correspondingly recorded in the URL category library. At this time, the entry corresponding to the URL in the URL category library includes the URL, the predicted category of the URL, the predicted probability, and the upper-level URL of the URL. For the lowest level URL, its predicted category and predicted probability are determined by the existing web page classification method, while the predicted category and predicted probability of URLs in other levels are determined according to the predicted category and predicted probability of the next level URL of this URL.

Specifically, determining the predicted category and predicted probability of the URL at the upper level according to the predicted category and predicted probability of the URL at the next level may be specifically implemented in the following manner:

Obtain all URLs whose upper level URL is the URL of the category and probability to be predicted from the URL category library; for each URL of the predicted category, calculate the weighted average of the predicted probability of each URL in the predicted category; The predicted category with the highest weighted average value is determined as the predicted category of the URL to be predicted, and the average value of predicted probabilities of URLs of the predicted category is determined as the predicted probability of the URL to be predicted.

Still taking the above four URLs as an example, assume that the prediction probabilities of the above four URLs are 80%, 79%, 81% and 80% in sequence. Then among the 4 URLs, the weighted average of the prediction probability of URLs whose prediction category is historical is 60% ((80%+79%+81%)/(3+1)), and the prediction of URLs whose prediction category is current commentary The weighted average of the probabilities is 20% ((80%)/(3+1)). Therefore, the predicted category of the URLs in the upper layer adjacent to the four URLs is history, and its predicted probability is 60%.

The above process can be realized by computer programs, or the URL category library can be manually configured according to the above principles.

It should be recognized that in the technical solution of the embodiment of the present invention, when the URL of the webpage to be classified is not recorded in the URL category library, it is not limited to determining the category of the webpage to be classified by querying the URL of the upper level of the URL layer by layer. The category of the webpage to be classified may also be determined by directly inquiring the predicted category of the URL of the URL of the previous level of the URL or other URLs of the URL of the upper level of the URL. In addition, in the technical solution provided by the embodiment of the present invention, the method for determining the URL prediction category of the upper level is not limited to the method described in the above process, and may also be other methods.

Through the above process, the upper-level URL of the URL recorded in the existing URL category library can be determined, and the entry corresponding to the upper-level URL can be stored in the URL category library. The entries stored in the URL category library can form a multi-level structure . Wherein, the data structure of the URL information in the updated URL category library can be shown in Table 2:

Table 2

name note url URL url_label Prediction category

prediction predicted probability faur level Upper level URL

Among them, the meaning of each variable is as follows:

url: URLStringUTF-8 of the web page

url_label: predicted category of URL StringUTF-8

prediction: URL prediction probability Double

faurlevel: upper level URLStringUTF-8

Based on the above-mentioned URL category library, the embodiment of the present invention provides a method for classifying webpages, as shown in FIG. 2 , which is a schematic diagram of the process flow of the method for classifying webpages provided by the embodiment of the present invention, which may include the following steps:

Step 201, obtain the URL of the webpage to be classified, and check whether the URL is recorded in the URL category library.

Step 202 , if it is found that the same URL is recorded in the URL category library, go to step 204 ; otherwise, go to step 203 .

Step 203 : Generate the upper-level URL of the URL, check whether the upper-level URL is recorded in the URL category library, and go to step 202 .

Step 204: Determine the predicted category corresponding to the queried URL as the category of the webpage to be classified.

Specifically, in the prior art solution, an exact match query is performed in the URL category library directly according to the URL, and when the corresponding table item is queried, the predicted category of the URL is returned; when the corresponding table item is not queried , returns a null value.

However, in the technical solution provided by the embodiment of the present invention, URLs are hierarchically divided by introducing, and entries corresponding to upper-level URLs are stored in the URL category library. When it is necessary to classify the webpage, firstly perform an exact match in the URL category library according to the URL of the webpage to be classified, and when the table item corresponding to the URL of the webpage to be classified is not stored in the URL category library, further generate the URL of the webpage to be classified The upper-level URL, and query the corresponding entry in the category library according to the upper-level URL, and use the predicted category of the queried upper-level URL as the predicted category of the URL of the webpage to be classified.

For example, the obtained URL of the webpage to be classified is http://sports.sina.com.cn/k/2011-05-18/09415581512.shtml, and there is no record related to the webpage to be classified in the current URL category database. URL, at this time, it is necessary to generate the upper-level URL of this URL, that is, http://sports.sina.com.cn/k/2011-05-18/, and query the corresponding upper-level URL in the URL category library table entry. If the entry corresponding to the upper-level URL is stored in the URL category library, the predicted category (such as sports) of the upper-level URL can be obtained by querying the URL category library, and the predicted category of the upper-level URL is used as the Classify the predicted category of URLs for web pages.

It should be noted that when the highest-level URL corresponding to the URL of the webpage to be classified has been found, but the same URL is not found in the URL category library, a query failure response is returned.

In the embodiment of the present invention, when a new lowest-level URL is added in the URL category library, the URL category library may be updated by means of event triggering or manual triggering. Specifically, the lowest-level URLs stored in the URL category library may be re-traversed, and hierarchically divided, and the corresponding upper-level URLs and corresponding predicted categories may be reacquired. In addition, it is also possible to only update the predicted categories of the upper-level URLs related to the newly added lowest-level URLs. The specific implementation will not be repeated here.

Based on the same technical idea as the above-mentioned webpage classification method, the embodiment of the present invention also provides a webpage classification device, which can be applied to the above-mentioned webpage classification method based on the URL category library. The URL category library records URLs of various levels, wherein, The upper-level URL among the URLs of the adjacent levels is intercepted and obtained on the basis of the lower-level URL, and each URL is recorded with a corresponding prediction category.

As shown in FIG. 3, a schematic structural diagram of a web page classification device provided in an embodiment of the present invention may include:

Upper-level URL generating module 31, for generating the upper-level URL of the URL according to the URL of the webpage to be classified;

Inquiry module 32, is used for according to the URL query URL classification storehouse of webpage to be classified; If do not inquire about matching URL, then query URL category storehouse according to the upper level URL of this URL;

The determination module 33 is configured to determine the category of the webpage to be classified according to the predicted category of the queried URL when the query module 32 queries a matching URL.

Wherein, the upper-level URL generation module 31 is specifically used to generate the upper-level URL of the URL when the query module 32 does not find a matching URL;

Inquiry module 32 specifically inquires the prediction category of the upper-level URL of the URL of the webpage to be classified by the following process:

Step A. Obtain the upper-level URL of the URL, and check whether the upper-level URL is recorded in the URL category library;

Step B. If the same URL is recorded in the URL category library, go to step C; otherwise, go to step A;

Step C, obtaining the predicted category of the queried URL;

The determining module 33 is specifically configured to determine the URL prediction category queried by the query module 33 as the category of the webpage to be classified.

Wherein, the determination module 33 is also used to return a query failure response when the query module 32 has already found the highest-level URL corresponding to the URL of the webpage to be classified, but has not yet found the same URL recorded in the URL category library.

Wherein, the webpage classification device also includes: URL category library maintenance module 34;

The upper-level URL generation module 31 is specifically used to traverse the URLs in the URL category library, and when traversing to a URL, select the URL from the URL category library, and generate the last URL of the URL according to the selected URL. Hierarchy URL;

The query module 32 is specifically used to query the URL category library according to the upper-level URL generated by the upper-level URL generation module 31;

The URL category maintenance module 34 is configured to, when the query module 32 finds no matching URL, determine the predicted category of the upper-level URL, and record the upper-level URL and its predicted category in the URL category library.

Wherein, the URL category library maintenance module 34 is specifically configured to determine the predicted categories of URLs in other levels except the lowest level according to the predicted categories of URLs in the next level of URLs.

Wherein, the URL category library maintenance module 34 is specifically used to obtain all URLs whose upper-level URLs are URLs of the category to be predicted from the URL category library; determine the quantity of URLs of each predicted category in the obtained URLs; The predicted category with the largest number of URLs is determined as the predicted category of URLs of the category to be predicted.

Wherein, each URL in the URL category library also has a corresponding prediction probability;

The URL category library maintenance module 34 is specifically configured to determine the predicted categories and predicted probabilities of URLs at other levels except the lowest level according to the predicted categories and predicted probabilities of URLs at the next level of URLs.

Wherein, the URL category library maintenance module 34 is specifically used to obtain all URLs whose upper-level URLs are the URLs of the category and probability to be predicted from the URL category library; for the URL of each predicted category, calculate the predicted category The weighted average of the predicted probabilities of each URL in the URL; the predicted category with the highest weighted average is determined as the predicted category of the URL to be predicted, and the average value of the predicted probabilities of the URLs of the predicted category is determined as the URL to be predicted predicted probability.

Wherein, when a new URL is added in the URL category library,

The upper-level URL generation module 31 is also used to generate the upper-level URL of the URL;

The query module 32 is specifically used to query the URL category library according to the upper URL of the URL;

The URL category storehouse maintenance module 34 is specifically used for, if query module 32 inquires about matching URL, then update the predictive category of upper-level URL; If query module 32 does not inquire about matching URL, then record this upper-level URL and The corresponding predicted category.

Wherein, the upper-level URL generation module 31 is specifically configured to divide the URL into layers according to the delimiters in the URL, and use the fields before the preset number of delimiters from the last digit of the URL as the upper-level URL of the URL.

Through the description of the above embodiments, those skilled in the art can clearly understand that the present invention can be realized by means of software plus a necessary general-purpose hardware platform, and of course also by hardware, but in many cases the former is a better embodiment . Based on this understanding, the essence of the technical solution of the present invention or the part that contributes to the prior art can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions to make a A computer device (which may be a personal computer, a server, or a network device, etc.) executes the methods described in various embodiments of the present invention.

Those skilled in the art can understand that the drawing is only a schematic diagram of a preferred embodiment, and the modules or processes in the drawing are not necessarily necessary for implementing the present invention.

Those skilled in the art can understand that the modules in the device in the embodiment can be distributed in the device in the embodiment according to the description in the embodiment, and can also be changed and located in one or more devices different from the embodiment. The modules in the above embodiments can be combined into one module, and can also be further split into multiple sub-modules.

The serial numbers of the above embodiments of the present invention are for description only, and do not represent the advantages and disadvantages of the embodiments.

The above disclosures are only a few specific embodiments of the present invention, however, the present invention is not limited thereto, and any changes conceivable by those skilled in the art shall fall within the protection scope of the present invention.

Claims (14)

1. a webpage classification method, it is characterized in that, be applied to the webpage classification process that realizes based on Uniform Resource Locator URL category storehouse, the predictive category of each level URL and each URL is recorded in the described URL category storehouse, wherein, adjacent The upper-level URL in the hierarchical URL is intercepted on the basis of the lower-level URL, and the method includes: Query the URL category library according to the URL of the webpage to be classified; If no matching URL is found, then query the URL category library according to the upper URL of the URL, and when a matching URL is found, determine the category of the webpage to be classified according to the predicted category of the URL that is inquired; Wherein, the generating process of the URL category library includes: Traversing the URLs in the URL category library, and when traversing to a URL, selecting the URL from the URL category library, and generating the upper-level URL of the URL according to the selected URL; Judging whether the generated upper-level URL already exists in the URL category library, and when the upper-level URL does not exist in the URL category library, determine the predicted category of the upper-level URL, and use the upper-level URL Hierarchical URLs and their predicted categories are recorded in the URL category library. 2. The method according to claim 1, wherein said querying the URL category storehouse according to the upper-level URL of the URL comprises: Step A, generating the upper-level URL of the URL, and querying whether the upper-level URL is recorded in the URL category library; Step B. If the same URL is recorded in the URL category library, go to step C; otherwise, go to step A; Step C, obtaining the predicted category of the queried URL. 3. The method according to any one of claims 1-2, characterized in that, except for the URL at the lowest level, the predicted categories of URLs at other levels are determined according to the predicted categories of URLs at the next level of the URL. 4. The method according to claim 3, wherein, according to the predicted category of the URL of the next level, the predicted category of the URL of the upper level is determined, specifically: Obtaining all URLs whose upper-level URLs are URLs of the category to be predicted from the URL category library; Determine the quantity of URLs of each prediction category in the acquired URLs; The predicted category with the largest number of URLs is determined as the predicted category of URLs of the category to be predicted. 5. The method according to claim 3, characterized in that, each URL in the URL category storehouse also has a prediction probability respectively; Determine the predicted category and predicted probability of the URL at the upper level according to the predicted category and predicted probability of the URL at the next level, specifically: Obtain all URLs whose upper-level URLs are URLs of categories and probabilities to be predicted from the URL category library; For URLs of each prediction category, calculate the weighted average of the prediction probabilities of each URL in the prediction category; The predicted category with the highest weighted average value is determined as the predicted category of the URL to be predicted, and the average value of predicted probabilities of URLs of the predicted category is determined as the predicted probability of the URL to be predicted. 6. The method according to claim 1, wherein, when a new URL is added in the URL category storehouse, the upper-level URL of the URL is generated, and the URL category storehouse is inquired according to the upper-level URL of the URL, if If a matching URL is found, update the predicted category of the upper-level URL; if no matching URL is found, record the upper-level URL and the corresponding predicted category in the URL category library. 7. The method according to claim 1, characterized in that determining the upper level URL of the URL is specifically: The URL is hierarchically divided according to the delimiter in the URL, and the field before the delimiter of the preset number from the end of the URL is used as the URL of the upper level of the URL. 8. A webpage classification device, characterized in that, it is applied to the webpage classification process realized based on the Uniform Resource Locator URL category library, and the URL category library is recorded with the predicted categories of each level URL and each URL, wherein the adjacent The upper-level URL in the hierarchical URL is intercepted on the basis of the lower-level URL, and the equipment includes: The upper-level URL generation module is used to generate the upper-level URL of the URL according to the URL of the webpage to be classified; The query module is used to query the URL category library according to the URL of the webpage to be classified; if no matching URL is found, then query the URL category library according to the upper URL of the URL; A determining module, configured to determine the category of the webpage to be classified according to the predicted category of the queried URL when the query module inquires about a matching URL; Among them, it also includes: URL category library maintenance module; The upper-level URL generation module is specifically configured to traverse the URLs in the URL category library, and when a URL is traversed, select the URL from the URL category library, and generate the upper URL of the URL according to the selected URL. one-level URL; The query module is specifically used to query the URL category library according to the upper-level URL generated by the upper-level URL generation module; The URL category maintenance module is used to determine the predicted category of the upper-level URL when the query module does not find a matching URL, and record the upper-level URL and its predicted category in the URL category library middle. 9. The apparatus of claim 8, wherein The upper-level URL generation module is specifically used to generate an upper-level URL of the URL when the query module does not find a matching URL; Described inquiry module specifically inquires the predicted category of the upper URL of the URL of the webpage to be classified by the following process: Step A. Obtain the upper-level URL of the URL, and check whether the upper-level URL is recorded in the URL category library; Step B. If the same URL is recorded in the URL category library, go to step C; otherwise, go to step A; Step C, obtaining the predicted category of the queried URL; The determining module is specifically configured to determine the URL prediction category queried by the query module as the category of the webpage to be classified. 10. The device according to any one of claims 8-9, wherein the URL category library maintenance module is specifically configured to determine, according to the predicted category of URLs at the next level of URLs, URLs at other levels except the lowest level predicted category. 11. The device according to claim 10, wherein the URL category library maintenance module is specifically configured to obtain all URLs whose upper-level URLs are URLs of the category to be predicted from the URL category library; The number of URLs of each predicted category among the obtained URLs; the predicted category with the largest number of URLs is determined as the predicted category of URLs of the category to be predicted. 12. The device according to claim 10, characterized in that, each URL in the URL category library is also respectively corresponding to a prediction probability; The URL category library maintenance module is specifically used to obtain all URLs whose upper-level URLs are URLs of categories and probabilities to be predicted from the URL category library; The weighted average of the predicted probabilities of each URL; the predicted category with the highest weighted average is determined as the predicted category of the URL to be predicted, and the average value of the predicted probabilities of the URLs of the predicted category is determined as the predicted probability of the URL to be predicted . 13. The device according to claim 10, when a new URL is added in the URL category library, The upper-layer URL generation module is also used to generate an upper-layer URL of the URL; The query module is specifically used to query the URL category library according to the upper URL of the URL; The URL category library maintenance module is specifically used for, if the query module queries a matching URL, then updates the predicted category of the upper URL; if the query module does not query a matching URL, then records the URL in the URL category library. The upper-level URL and the corresponding prediction category. 14. The device according to claim 8, wherein the upper-layer URL generation module is specifically configured to divide the URL into layers according to the delimiter in the URL, and divide the URL by a preset number from the last bit forward The field before the delimiter is used as the upper level URL of this URL.