Method, device and equipment for improving search result based on user behaviors

[0023] The present invention will be described in further detail below in conjunction with the accompanying drawings.

[0024] figure 1 It is a flowchart of a method for improving search results based on user behaviors in one aspect of the present invention. Wherein, the method according to the present invention can be implemented by an operating system or a processing controller in a computer. For brevity, the operating system or processing controller is collectively referred to as a search device below. The computer includes but is not limited to: 1) user equipment; 2) network equipment. Wherein, the user equipment includes, but is not limited to: personal computers, smart phones, PDAs, etc.; the network equipment includes, but is not limited to: a single network server, a server group composed of multiple network servers, or cloud computing-based A cloud composed of a large number of computers or network servers. Among them, cloud computing is a kind of distributed computing, a super virtual computer composed of a group of loosely coupled computer sets.

[0025] In step S1, the search device acquires an input sequence from the user.

[0026] Wherein, the manner in which the search device obtains the input sequence includes but is not limited to:

[0027] 1) Obtain the input sequence from the user in real time;

[0028] Specifically, when the search device is included in the user equipment, the search device directly obtains the input sequence input by the user; when the search device is included in the network device, the search device obtains the input sequence input by the user in real time via the user equipment.

[0029] 2) Obtain the pre-stored input sequence from the user;

[0030] Specifically, the search device obtains a pre-stored input sequence, and the input sequence is obtained by storing the obtained input sequence from the user by a computer or other equipment.

[0031] Next, in step S2, the search device obtains candidate search result items and their true popularity corresponding to the input sequence according to the input sequence.

[0032] Specifically, the search device obtains candidate search result items according to the input sequence, and then, the search device obtains corresponding real popularity according to the candidate search result items and the input sequence. Wherein, each candidate search result item includes description information and/or acquisition information of a network resource or local resource, for example, includes link information and text description information of the network resource or local resource, or directly includes the resource itself. The network resources or local resources include, but are not limited to: 1) text information; 2) web page information; 3) multimedia information, such as picture information, audio information, and video information.

[0033] Wherein, the manner in which the search device obtains the candidate search result item includes but is not limited to:

[0034] 1) According to the input sequence, search in a preset inverted index library to obtain candidate search result items;

[0035] Specifically, the index term and the candidate search result item corresponding to the index term are preset in the inverted index library, and the search device matches the input sequence with the index term, and obtains the candidate corresponding to the successfully matched index term The search result item serves as the candidate search result item of the input sequence.

[0036] 2) According to the input sequence, perform a matching query in the description information and/or acquisition information corresponding to all candidate search result items to obtain candidate search result items;

[0037] Specifically, the search device queries the description information corresponding to each candidate search result item and/or whether the acquired information contains all or part of the information of the input sequence, and when the query obtains all or part of the information including the input sequence, the The candidate search result item serves as the candidate search result item of the input sequence.

[0038] Wherein, the method for the search device to obtain the true popularity includes but is not limited to:

[0039] 1) The search device directly obtains the preset real popularity corresponding to each candidate search result item and the input sequence.

[0040] Specifically, each candidate search result item corresponds to one or more real popularity, and each real popularity corresponds to an input sequence. The true popularity and its corresponding input sequence are stored in a computer device or other devices physically separated from the computer device but communicatively connected. After obtaining the candidate search result items of the input sequence, the search device queries computer equipment or other equipment and obtains the true popularity of the candidate search result items corresponding to the input sequence currently being processed.

[0041] For example, if the input sequence is "game download", the candidate search result items obtained by the search device according to the input sequence include the following candidate search result items:

[0042] " Stand-alone game_single-player game download_dote fun stand-alone game download base

[0043] Duoter stand-alone game download base provides stand-alone game players with the latest classic stand-alone game downloads, free download of popular and fun stand-alone games. Single-player game downloads, starting from the multi-special single-player game download base, shared by millions of people! "

[0044] The candidate search result item is preset with multiple real popular degrees corresponding to different input sequences, and the corresponding relationship between the input sequence and the real popular degrees is as follows:

[0045] Input sequence True popularity

[0046] Classic game 1.234

[0047] Game download 1.456

[0048] Stand-alone game 2.487

[0049] Dot 1.112

[0050] Then, the search device queries that the true popularity with the candidate search result item corresponding to the input sequence "game download" is 1.456.

[0051] 2) The search device obtains a plurality of pre-stored factors related to the real popularity, and obtains the real popularity according to the multiple factors:

[0052] Among them, multiple factors related to the real popularity include at least one of the following:

[0053] 1) The relative click rate of the candidate search result item corresponding to the input sequence;

[0054] Wherein, the relative click rate is obtained according to the number of clicks and the number of presentations of the candidate result item corresponding to the input sequence;

[0055] 2) The number of user recommendations corresponding to the candidate search result item and the input sequence;

[0056] 3) The number of user favorites corresponding to the candidate search result item and the input sequence.

[0057] The method of obtaining the true popularity based on the above factors will be described in detail in the subsequent content.

[0058] It should be noted that the above examples are only to better illustrate the solution of the present invention, rather than to limit the present invention. The real popularity can also be characterized by other ways, for example, using ranks for characterization, etc., any according to the input sequence The implementation of obtaining candidate search result items and their true popularity corresponding to the input sequence should all be included in the scope of the present invention.

[0059] Next, in step S3, the search device sorts the candidate search result items according to the true popularity to obtain a sort result.

[0060] Specifically, the search device sorts the candidate search result items corresponding to the real popularity in a descending manner of the real popularity. Among them, the way from high to low includes but is not limited to:

[0061] 1) When the true popularity is represented by a value, the candidate search result items are sorted according to the value of the true popularity from high to low;

[0062] 2) When the real popularity is represented by a rank, the candidate search result items are sorted according to the rank of the real popularity from high to low.

[0063] It should be noted that the above examples are only to better illustrate the solution of the present invention, rather than to limit the present invention. Any sorting of the candidate search result items according to the true popularity to obtain the realization of the ranking results All methods should be included in the scope of the present invention.

[0064] The method according to the present invention can be used to feed back the sorting result according to the user's input sequence in real-time online, and can also be used for offline processing to obtain the sorting result corresponding to the input sequence. Further, the method according to the present invention can be used to feed back the sorting results according to the query input sequence entered by the user in the search box, for example, by inputting the query input sequence in the search box provided by the search engine or website to obtain the result via the Internet Sort the results, or perform resource query through web pages or clients in the local area network; or, in the input method, provide users with sorted short texts according to the input sequence input by the user.

[0065] Since the method according to the present invention refers to the real popularity to determine the sorting result, it can better reflect the real concerns of the user compared with the prior art, so that the sorting result is more in line with the user's needs.

[0066] As one of the preferred solutions of the present invention, the present invention also includes step S4 (not shown in the figure). In step S4, the search device determines the true popularity corresponding to the candidate search result item under the input sequence according to at least one of the following:

[0067] 1) The relative click rate of the candidate search result item corresponding to the input sequence;

[0068] Specifically, the search device obtains the relative click-through rate according to the recorded number of clicks and display times of the candidate search result item corresponding to the input sequence.

[0069] For example, the search device is based on the formula:

[0070] Relative click rate=k*(number of clicks/number of impressions) to obtain relative click rate;

[0071] Among them, k is a parameter that can be determined by a person skilled in the art according to actual conditions and requirements.

[0072] For another example, the search device according to the formula:

[0073] Relative click rate = clicks/(number of impressions-clicks) to obtain the relative click rate.

[0074] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any item corresponding to the input sequence according to the candidate search result The click-through rate to obtain the real popularity should be included in the scope of the present invention.

[0075] Wherein, the number of clicks indicates the number of times the user clicks on the candidate search result item under the input sequence, and the number of presentations indicates the number of times the candidate search result item is presented to the user under the input sequence.

[0076] Specifically, the search device or other device obtains the number of times of display by recording the number of times the candidate search result item is displayed in front of the user under each input sequence. When the number of presentations needs to be processed, the search device queries the recorded corresponding number of presentations according to the input sequence and candidate search result items.

[0077] Among them, the ways to judge the candidate search result items to be displayed in front of the user include but are not limited to:

[0078] i) Determine that the candidate search result item presented through the browser is the candidate search result item displayed in front of the user;

[0079] The search device or other devices provide the browser with data for displaying to the user directly or via the user equipment according to the user's confirmation of the search behavior and the input sequence input by the user. Then, the search device or other devices determine the candidate search result items included in the data provided to the browser as candidate search result items displayed in front of the user under the input sequence.

[0080] For example, the user accesses the Baidu search engine through the user device, enters the input sequence "game download" and clicks the button "Baidu Click" to confirm the search. Then, the Baidu search engine determines that the candidate search result items included in the data fed back to the user equipment according to the input sequence "game download" are candidate search result items displayed in front of the user under the input sequence "game download".

[0081] ii) Determine that the candidate search result item presented by the client is the candidate search result item presented in front of the user;

[0082] The search device or other device directly or via the user equipment provides the client with data for presenting to the user according to the user's behavior. Then the search device or other devices determine the candidate search result items included in the data provided to the client as candidate search result items displayed in front of the user.

[0083] For example, a user accesses a local search engine of a paper library through a paper search client, enters the input sequence "game download" and clicks the button "confirm" to confirm the search. Then the local search engine judges that the candidate search result items contained in the data fed back to the client according to the input sequence "game download" are candidate search result items displayed in front of the user under the input sequence "game download".

[0084] iii) Determine that the displayed input method candidates are candidate search result items displayed in front of the user;

[0085] When the candidate search result item is an input method candidate, the search device or other device judges the input method candidate included in the data used to generate the input method candidate column as the candidate search result displayed in front of the user under the input sequence item.

[0086] For example, the input sequence input by the user is "youxi", and the input method candidates included in the data used to generate the input method candidate column obtained by the user device based on the input sequence are "game", "youxi", and "youxi" , "Youxi" and "youxi", the search device or other device determines that the above five input method options are the candidate search result items displayed in front of the user under the input sequence "youxi". It should be noted that when the user performs the operation of presenting the next page, for example, selects the "pagedown" button, etc., the input method candidates included in the data in the update input method candidate column are also judged as being in the input sequence. Candidate search result items displayed in front of users under "youxi".

[0087] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any number of candidate search result items displayed in front of the user by recording each input sequence All the ways to achieve the number of times of presentation should be included in the scope of the present invention.

[0088] The search device or other devices also obtain the number of clicks by the user by recording the number of times each candidate search result item under each input sequence is clicked. When the number of clicks needs to be processed, the search device queries the recorded corresponding number of clicks according to the input sequence and candidate search result items. The search device or other device uses the number of times selected by the user as the number of clicks during the process of displaying the candidate search result items, for example, during the process of displaying the candidate search result items through a browser, client or input method candidate column. Those skilled in the art should be able to determine the implementation manner of obtaining the number of clicks, which will not be repeated here.

[0089] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any click corresponding to the input sequence according to the candidate search result item The number of times and the number of times of presentation to achieve the relative click rate should be included in the scope of the present invention.

[0090] 2) The number of user recommendations corresponding to the candidate search result item and the input sequence;

[0091] Specifically, when a device providing candidate search result items provides a function of recommending each candidate search result item to tourists or registered users, the device records the number of recommendations for each candidate search result item obtained based on the input sequence and recommended by the user. When the number of recommendations needs to be processed, the search device queries the number of recommendations recorded according to the input sequence and candidate search result items, and obtains the true popularity according to the number of recommendations, for example, directly use the number of recommendations as the real Popularity, or after processing the recommended number, for example, normalizing or dividing by a predetermined coefficient, the processed value is used as the true popularity.

[0092] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any user who corresponds to the input sequence according to the candidate search result item The number of recommendations to obtain the real popularity should be included in the scope of the present invention.

[0093] 3) The number of user favorites corresponding to the candidate search result item and the input sequence;

[0094] Specifically, when the device that provides candidate search result items provides a function for registered users to bookmark each candidate search result item, the device records the number of favorite search result items obtained by the user based on the input sequence. When the number of favorites needs to be processed, the search device queries the recorded number of favorites according to the input sequence and candidate search result items, and obtains the true popularity based on the number of favorites, for example, directly regards the number of favorites as the true Popularity, or after processing the number of favorites, for example, normalizing or dividing by a predetermined coefficient, the processed value is used as the true popularity.

[0095] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any user who corresponds to the input sequence according to the candidate search result item The number of favorites to obtain the real popularity should be included in the scope of the present invention.

[0096] It should be noted that the search device may also combine any number of the relative click rate, the number of user recommendations, and the number of user favorites to obtain the true popularity. For example, the search device normalizes the relative click-through rate, the number of user recommendations, and the number of user favorites respectively, and then obtains the average value or sum of the three above or performs corresponding processing according to a preset formula to obtain the true popularity. Those skilled in the art should understand that the scope of the present invention is not limited to the foregoing examples.

[0097] It should be further explained that the method of determining the true popularity described in this embodiment can also be used in step S3, the search device obtains a plurality of pre-stored factors related to the true popularity, and according to the multiple factors Factors to obtain the true popularity.

[0098] As one of the preferred solutions of the present invention, in step S3, the search device further sorts the candidate search result items based on the first predetermined rule and according to the true popularity to obtain the sort result.

[0099] Specifically, the first predetermined rule includes a sorting rule for determining the sorting result of the candidate search result item based on one or more existing factors in combination with the true popularity. Then the search device obtains the one or more existing factors according to the sorting rule specified by the first predetermined rule, and determines the candidate search result item based on the obtained existing factors and the true popularity. Sort the results.

[0100] Among them, the existing factors include but are not limited to:

[0101] 1) The degree of correlation between the input sequence and the candidate search result item;

[0102] The correlation degree may be obtained according to factors such as the degree of matching between the input sequence and the candidate search result item, the category of the input sequence, and/or the category of the candidate search result item;

[0103] 2) The importance of the candidate search result item;

[0104] The importance can be obtained according to factors such as click-through rate and authority of candidate search result items.

[0105] Since the above two factors are all existing factors, those skilled in the art should be able to obtain the above two factors based on the input sequence and the candidate search result items according to actual conditions and needs. Therefore, the two factors are not mentioned here. The method of obtaining the factors is described in detail.

[0106] It should be noted that the above-mentioned examples of existing factors are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any other existing factors that affect the ranking of candidate search results All factors should be included in the scope of the existing factors described in the present invention.

[0107] The sorting rules further include any of the following rules:

[0108] 1) First determine the initial ranking result of the candidate search result items according to existing factors, and then adjust the initial ranking result according to the true popularity to obtain the rules of the ranking result;

[0109] For example, for the input sequence "game download", the initial ranking result of candidate search result items obtained by the search device according to existing factors is:

[0110] Candidate search result item one;

[0111] Candidate search result item three;

[0112] Candidate search result item four;

[0113] Candidate search result item two;

[0114] Candidate search result item five.

[0115] The true popularity of each candidate search result item obtained by the search device and the input sequence "Game Download" is as follows:

[0116] Candidate search result item True popularity

[0117] Candidate search result item one first level

[0118] Candidate search result item 2 Level 1

[0119] Candidate search result item 3 Level 3

[0120] Candidate search result item 4 Second level

[0121] Candidate search result item five third level

[0122] If the sorting rule includes analyzing the true popularity of two adjacent candidate search result items from front to back, when the true popularity level of the candidate search result item is higher than that of the previous candidate search result item, the two positions will be mutually Change; when the true popularity level of the candidate search result item in the back is lower than or equal to the candidate search result item in the front, the rules of the two ranking positions are not changed.

[0123] According to the above rules, the search device analyzes that the true popularity level of candidate search result item 1 is higher than that of candidate search result item three, and the ranking positions of the two are not changed; then, the search device analyzes and obtains candidate search result item three The true popularity level of the candidate search result item 4 is lower than the true popularity level of the candidate search result item four, the positions of the candidate search result item 3 and the candidate search result item 4 are swapped; then, because the position of the candidate search result item 3 has changed from the original The second place is changed to the third place. Therefore, the search device compares the real popularity level of candidate search result item 3 with the real popularity level of candidate search result item 2, and judges the true popularity level of candidate search result item 3. If it is lower than the true popularity level of candidate search result item 2, the positions of candidate search result item 3 and candidate search result item 2 are swapped; then, because the position of candidate search result item 3 has changed from third to fourth Therefore, the search device compares the true popularity level of candidate search result item three with the true popularity level of candidate search result item five, and determines that the true popularity level of candidate search result item three is equal to that of candidate search result item five. The real popularity level does not change the ranking position of the two. After the search device performs the above operations, the result of the candidate search term ranking obtained is:

[0124] Candidate search result item one;

[0125] Candidate search result item four;

[0126] Candidate search result item two;

[0127] Candidate search result item three;

[0128] Candidate search result item five.

[0129] 2) Synthesize the existing factors and the real popularity to directly obtain the rules of the ranking result.

[0130] For another example, for the input sequence "game download", the existing factors referred to by the sorting rule include the relevance of the input sequence and the candidate search result item and the importance of the candidate search result item, and the ranking The rule stipulates that the relevance, importance, and true popularity are added, and the candidate search result items are sorted according to the sum obtained. If the relevance and importance obtained by the search device in step S3 and the real popularity obtained by the search device in step S2 are as follows:

[0131] Candidate search results Item relevance Importance True popularity

[0132] Candidate search result item one 2.135 2.167 1.234

[0133] Candidate search result item two 3.214 3.246 1.456

[0134] Candidate search result item three 2.489 2.498 2.487

[0135] Candidate search result item four 1.358 1.397 1.112

[0136] Candidate search result item five 1.698 2.054 1.587

[0137] Then, the search device calculates the sum of the relevance, importance, and true popularity of each candidate search result item according to the regulations of the sorting rule, and sorts the candidate search result items according to the added sum, and the sorting result obtained is:

[0138] Candidate search result item two;

[0139] Candidate search result item three;

[0140] Candidate search result item one;

[0141] Candidate search result item five;

[0142] Candidate search result item four.

[0143] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any method based on the first predetermined rule and according to the actual popularity will be The candidate search result items are sorted to obtain the realization method of the sorting result, for example, the rule of adjusting the initial sorting result according to the real popularity is changed, or, when the real popularity is represented in the form of value, according to the real popularity To adjust the initial ranking result, or to weight each existing factor and real popularity, and then multiply or average the weighted existing factors and real popularity to obtain the All sorting values ​​should be included in the scope of the present invention.

[0144] As one of the preferred solutions of the present invention, step S2 further includes step S21 (not shown) and step S22 (not shown).

[0145] In step S21, the search device acquires the information units included in the input sequence.

[0146] Specifically, the method of obtaining the information unit includes but is not limited to:

[0147] 1) The search device segments the input sequence to obtain a word segmentation unit as an information unit;

[0148] 2) The search device extracts the n-gram segment of the input sequence to use the n-gram segment as an information unit.

[0149] Furthermore, the invalid information in the word segmentation unit or n-gram segment obtained by the above two methods can be removed, so that the remaining word segmentation unit or n-gram segment can be used as the information unit.

[0150] Next, in step S22, the search device obtains the candidate search result item according to the information unit.

[0151] Specifically, the search device first obtains candidate search result items according to the information unit, and then uses the candidate search result items obtained based on the information unit as the candidate search result items of the input sequence to which the information unit belongs. The manner in which the search device obtains candidate search result items according to the information unit is the same or similar to the manner in which the search device obtains candidate search result items according to the input sequence in step S2, and will not be repeated here.

[0152] As one of the preferred solutions of the present invention, the method according to the present invention further includes step S5 (not shown), step S6 (not shown), and step S7 (not shown).

[0153] In step S5, the search device establishes or updates the ranking index information according to the input sequence and the ranking result of the corresponding candidate search result items.

[0154] Wherein, the sort index information includes any one of the following index information:

[0155] 1) The input sequence, the candidate search result item corresponding to the input sequence, and the ranking position of the candidate search result item under the input sequence;

[0156] Specifically, the search device directly establishes or updates the input sequence, the candidate search result item corresponding to the input sequence, and the index information of the ranking position of the candidate search result item under the input sequence according to the sorting result obtained in step S3.

[0157] 2) The input sequence, the candidate search result item corresponding to the input sequence, and the ranking position adjustment information of the candidate search result item under the input sequence;

[0158] Specifically, the search device obtains the ranking position adjustment information of the candidate search result item under the input sequence according to the real popularity or the rank or weight value of other existing factors contained in the ranking result obtained in step S3, for example, weighting Information or weight reduction information, etc., to establish or update the input sequence, the candidate search result item corresponding to the input sequence, and the index information of the position adjustment information of the candidate search result item sorted under the input sequence.

[0159] In step S6, the search device acquires the input sequence from the user in real time.

[0160] Next, in step S7, the search device performs a matching query in the ranking index information according to the input sequence acquired in real time to obtain a ranking result corresponding to the input sequence acquired in real time.

[0161] Specifically, the search device directly queries and obtains the ranking result of the input sequence according to the input sequence contained in the ranking index information, the candidate search result item corresponding to the input sequence, and the ranking position of the candidate search result item under the input sequence. .

[0162] Alternatively, the search device processes the candidate search result items according to the input sequence contained in the ranking index information, the candidate search result item corresponding to the input sequence, and the ranking position adjustment information of the candidate search result item under the input sequence to Get the sort result of the input sequence.

[0163] Preferably, the present invention further includes the step of selecting a search result item for presentation to the user from the candidate search result items according to the ranking result.

[0164] Specifically, the search device selects the search result items to be presented to the user from the candidate search result items according to the user behavior and the number of candidate search result items that need to be presented.

[0165] figure 2 It is a flowchart of a method for improving search results based on user behavior in a preferred embodiment of the present invention. The method according to the present invention includes step S1, step S2 and step S31.

[0166] Step S1 and Step S2 are already referencing figure 1 The illustrated embodiment is described in detail, and is included here by reference, and will not be repeated.

[0167] In step S31, the search device sorts the candidate search result items based on the first predetermined rule and according to the real popularity and other sorting-related information to obtain the sorting result.

[0168] The other sorting related information includes at least one of the following:

[0169] 1) Other user behavior information related to the candidate search result item;

[0170] Wherein, the other user behavior information can be obtained by a search device or other devices, and the search device can obtain the other user behavior information obtained by other devices through a communication connection or other connection methods, and the other user behavior information includes at least one of the following:

[0171] i) The presentation time of the candidate search result item corresponding to the input sequence;

[0172] The presentation time indicates the time that the candidate search result item is displayed in front of the user. Specifically, when the candidate search result item is displayed through the browser, the display time represents the time from when the browser generates the page to when the user updates the candidate search result item on the page or closes the page; when the candidate search result item passes the client When the client is displayed, the display time represents the time from when the client generates the display page to when the user updates the candidate search result item in the display page or closes the page; when the candidate search result item is displayed through the input method candidate column, the display time Represents the time when the candidate search result item is displayed in the input method candidate column.

[0173] ii) The display position of the candidate search result item on the page;

[0174] The display position includes the position of the candidate search result item in the web page, client page, or input method candidate column when it is displayed to the user. For example, for web pages and client pages, the display position includes upper, left, right, lower, middle, title, text, sidebar, etc. For input method candidate columns, the display position includes input How many options are in the candidate column?

[0175] iii) The browsing time corresponding to the resource corresponding to the candidate search result item and the input sequence;

[0176] The browsing time indicates the length of time for the user to browse the candidate search result item after selecting it. For example, for a browser or client to present a candidate search result item, the browsing time can be based on the time for browsing the resource after the user opens the resource corresponding to the candidate search result item through a click or keyboard selection behavior; where The time when the resource is displayed is used to obtain the browsing time. For example, if the resource is a webpage, the obtained time of existence of the webpage is used as the user's browsing time.

[0177] Specifically, the search device adjusts the ranking result based on the first predetermined rule according to at least one item of other user behavior information, and the ranking result obtained according to the real popularity; or, the search device directly based on the existing existing rules referenced by the first predetermined rule. Factors, real popularity, and other user behavior information to obtain the ranking results.

[0178] Among them, the longer the candidate search result item is displayed, the higher the ranking; the farther the candidate search result item is displayed, and the farther it is selected by the user, the higher the ranking. It is assumed that, for example, the middle part and title of the preset page, the first item in the input method candidate column is the user's attention area, etc.; the longer the browsing time of the candidate search result item, the higher the ranking. The search device can determine the adjustment level to which the display time, the display location, and the browsing time belong according to any one or several items of the obtained information of the display time, the display location, and the browsing time, and determine the adjustment level according to the adjustment level. Adjust the aforementioned sorting results, the way to adjust the sorting results according to the adjustment level, and the reference figure 1 In the illustrated embodiment, the method of adjusting the initial ranking results according to the actual popularity is the same or similar, and is included here by reference, and will not be repeated; or, the search device is based on the obtained display time, display position, and browsing time. Any item or any number of items in the information is used to determine the individual weight value corresponding to the display time, display position, and browsing time, and to obtain the sorting result according to the individual weight value. This method will be described in detail in subsequent embodiments. I will not repeat them here.

[0179] 2) User related information;

[0180] Among them, the user's identity can be identified according to the user's registration information; the identification code of the user's device, such as mobile phone number or hardware hash value, etc.; and the user's temporary record information, such as user cookie. The user-related information includes at least one of the following:

[0181] i) Personal attributes of the user;

[0182] The personal attributes include, but are not limited to, the age, gender, occupation, income, education level and other information of the user actively provided by the registered user or inferred from the recorded user behavior. The search device adjusts the sorting results according to the user's personal attributes. For example, if the user is a female, the candidate search result items whose category is "shopping" are sorted in advance. Among them, the category of the candidate search result item is obtained through preset information.

[0183] ii) User preferences;

[0184] The search device adjusts the sorting result according to the user's preference setting. For example, if the preference setting includes a setting to advance candidate search result items of the "military" category. Then, the search device advances the sorting position of the candidate search result items of the "military" category in the sorting result according to the preference setting.

[0185] iii) User's environmental information;

[0186] The environmental information includes time information and location information. Wherein, the time information may be obtained through the current time of the user equipment, and the location information may be obtained according to the current location information of the user equipment, for example, IP address or mobile phone roaming area. The search device adjusts the ranking result according to the environmental information. For example, if the search device analyzes that the user's current IP address is Shanghai, and the user input sequence is "weather forecast", the weather forecast ranking position in the Shanghai area will be advanced, etc.

[0187] iv) User's historical behavior record;

[0188] The user's historical behavior record includes but is not limited to: user's click behavior, user's browsing behavior, etc. The search device adjusts the sorting results according to the user's historical behavior records. For example, if the historical behavior record includes the record of the user selecting the weather forecast in the Shanghai area when inputting the sequence "weather forecast", when the user's input sequence is obtained as "weather forecast", the weather forecast in the Shanghai area will be sorted. Wait in advance.

[0189] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any based on the first predetermined rule, according to the real popularity and other ranking related Information to sort the candidate search result items to obtain the sorting result should all be included in the scope of the present invention.

[0190] image 3 It is a flowchart of a method for improving search results based on user behavior in another preferred embodiment of the present invention. The method according to this embodiment includes step S1, step S2, and step S3. Among them, step S3 further includes step S32 and step S33.

[0191] Step S1 and Step S2 are already referencing figure 1 It is described in detail in the illustrated embodiment, and is included here by reference, and will not be repeated.

[0192] In this embodiment, the first predetermined rule further includes a weight ordering rule.

[0193] In step S32, the search device obtains the weight value of the candidate search result item under the input sequence based on the weight ranking rule, according to the real popularity and the other ranking related information.

[0194] Specifically, the weight ranking rule refers to existing factors, real popularity, and other related ranking information to determine the weight value. Among them, the existing factors, real popularity and other ranking related information have been referenced figure 1 or figure 2 It is described in detail in the illustrated embodiment, and is included here by reference, and will not be repeated. The weight ordering rule further includes any one of the following rules:

[0195] 1) The individual weight value corresponding to any one of the existing factors, real popularity or other ranking related information is used as the initial weight value, and then the initial weight value is adjusted according to the individual weight values ​​corresponding to the other two to obtain the aforementioned Weights.

[0196] For example, according to the weight ranking rule, the search device uses the individual weight value corresponding to the existing factor as the initial weight value, and then, according to the difference between the individual weight value corresponding to the true popularity and other ranking related information and the individual weight value corresponding to the existing factor The predetermined range to which the value belongs is used to determine the upward or downward adjustment range of the initial weight value. Among them, those skilled in the art should be able to determine the predetermined range according to actual conditions and requirements. For example, if the individual weight values ​​corresponding to the true popularity and other ranking related information are greater than the individual weight values ​​corresponding to the existing factors, and the difference falls within a predetermined range between 0.2 and 0.4, the initial weight value is increased by 0.1, etc.

[0197] Among them, those skilled in the art should be able to determine the individual weight value of the existing factor according to the actual situation, which will not be repeated here.

[0198] The individual weight value corresponding to the true popularity is based on the relative click rate of the candidate search result item and the input sequence, the number of user recommendations corresponding to the candidate search result item and the input sequence, and the candidate search result The item is obtained from at least one of the user favorites corresponding to the input sequence. For example, the search device directly uses the value corresponding to the relative click rate, the number of recommendations, or the number of favorites as the weight value of the real popularity; another example, the relative click rate, the number of recommendations, and the number of favorites are respectively normalized and added, and The obtained sum serves as the weight value corresponding to the real popularity and so on. Those skilled in the art should understand that the method of obtaining the individual weight value corresponding to the real popularity is not limited to the above example.

[0199] The individual weight values ​​corresponding to other ranking related information are obtained according to one or more types of information contained in the other ranking related information. For example, the search device directly uses any one of the corresponding values ​​in other user behavior information as the weight value of other ranking-related information; for another example, the search device normalizes the values ​​corresponding to each item of other user behavior information. , And obtain the average value of each normalized value as the individual weight value corresponding to other sorting related information. Those skilled in the art should understand that the method for obtaining the individual weight values ​​corresponding to other sorting related information is not limited to the above examples.

[0200] 2) Obtain the aforementioned weight value directly according to the individual weight value corresponding to at least one of the existing factors, real popularity or other ranking related information.

[0201] For example, the weight ranking rules stipulate that the weighting coefficients of existing factors, real popularity or other ranking-related information are 0.8, 1.2, and 1.1 respectively, and stipulate that the weight values ​​of each individual item are multiplied by the corresponding weighting coefficients to obtain the weight value. Then, the search device multiplies the weight values ​​of the individual items by the corresponding weight coefficients and adds them, and uses the sum obtained by the addition as the weight value.

[0202] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any sorting rules based on the weights are based on the true popularity and the Other sorting-related information to obtain the weight value of the candidate search result item under the input sequence, such as using other weight adjustment methods or other weight calculation methods, should be included in the scope of the present invention .

[0203] In step S33, the search device sorts the candidate search result items according to the weight value to obtain the sort result.

[0204] Specifically, the search device sorts the candidate search result items in a descending order of the weight value to obtain the sort result.

[0205] Figure 4 It is a schematic structural diagram of a search device for improving search results based on user behavior in one aspect of the present invention. The search device according to the present invention includes a first acquiring device 1, a second acquiring device 2, and a sorting device 3.

[0206] The first obtaining device 1 obtains an input sequence from a user.

[0207] Wherein, the manner in which the first obtaining device 1 obtains the input sequence includes but is not limited to:

[0208] 1) Obtain the input sequence from the user in real time;

[0209] Specifically, when the first acquiring device 1 is included in the user equipment, the first acquiring device 1 directly acquires the input sequence input by the user; when the first acquiring device 1 is included in the network device, the first acquiring device 1 Obtain the input sequence input by the user in real time via the user device.

[0210] 2) Obtain the pre-stored input sequence from the user;

[0211] Specifically, the first acquiring device 1 acquires a pre-stored input sequence, and the input sequence is obtained by storing the acquired input sequence from the user by a computer or other equipment.

[0212] The second obtaining device 2 obtains candidate search result items and their true popularity corresponding to the input sequence according to the input sequence.

[0213] Specifically, the second obtaining device 2 obtains candidate search result items according to the input sequence, and then, the second obtaining device 2 obtains corresponding real popularity according to the candidate search result items and the input sequence. Wherein, each candidate search result item includes description information and/or acquisition information of a network resource or local resource, for example, includes link information and text description information of the network resource or local resource, or directly includes the resource itself. The network resources or local resources include, but are not limited to: 1) text information; 2) web page information; 3) multimedia information, such as picture information, audio information, and video information.

[0214] Wherein, the manner in which the second obtaining device 2 obtains the candidate search result item includes but is not limited to:

[0215] 1) According to the input sequence, search in a preset inverted index library to obtain candidate search result items;

[0216] Specifically, the index words and candidate search result items corresponding to the index words are preset in the inverted index library, and the second obtaining device 2 matches the input sequence with the index words, and obtains the index words that are successfully matched The corresponding candidate search result item is used as the candidate search result item of the input sequence.

[0217] 2) According to the input sequence, perform a matching query in the description information and/or acquisition information corresponding to all candidate search result items to obtain candidate search result items;

[0218] Specifically, the second obtaining device 2 queries the description information corresponding to each candidate search result item and/or whether the obtained information contains all or part of the information of the input sequence, and when the query obtains all or part of the information including the input sequence To use the candidate search result item as the candidate search result item of the input sequence.

[0219] Wherein, the way for the second acquiring device 2 to acquire the true popularity includes but is not limited to:

[0220] 1) The second acquiring device 2 directly acquires the preset real popularity corresponding to each candidate search result item and the input sequence.

[0221] Specifically, each candidate search result item corresponds to one or more real popularity, and each real popularity corresponds to an input sequence. The true popularity and its corresponding input sequence are stored in a computer device or other devices physically separated from the computer device but communicatively connected. After obtaining the candidate search result item of the input sequence, the second obtaining device 2 queries and obtains the true popularity of the candidate search result item corresponding to the currently processed input sequence in a computer device or other equipment.

[0222] For example, if the input sequence is "game download", the candidate search result items obtained by the second obtaining device 2 according to the input sequence include the following candidate search result items:

[0223] " Stand-alone game_single-player game download_dote fun stand-alone game download base

[0224] Duoter stand-alone game download base provides stand-alone game players with the latest classic stand-alone game downloads, free download of popular and fun stand-alone games. Single-player game downloads, starting from the multi-special single-player game download base, shared by millions of people! "

[0225] The candidate search result item is preset with multiple real popular degrees corresponding to different input sequences, and the corresponding relationship between the input sequence and the real popular degrees is as follows:

[0226] Input sequence True popularity

[0227] Classic game 1.234

[0228] Game download 1.456

[0229] Stand-alone game 2.487

[0230] Dot 1.112

[0231] Then, the second acquiring device 2 inquires that the real popularity with the candidate search result item corresponding to the input sequence "game download" is 1.456.

[0232] 2) The second obtaining device 2 obtains a plurality of pre-stored factors related to the real popularity, and obtains the real popularity according to the multiple factors:

[0233] Among them, multiple factors related to the real popularity include at least one of the following:

[0234] 1) The relative click rate of the candidate search result item corresponding to the input sequence;

[0235] Wherein, the relative click rate is obtained according to the number of clicks and the number of presentations of the candidate result item corresponding to the input sequence;

[0236] 2) The number of user recommendations corresponding to the candidate search result item and the input sequence;

[0237] 3) The number of user favorites corresponding to the candidate search result item and the input sequence.

[0238] The method of obtaining the true popularity based on the above factors will be described in detail in the subsequent content.

[0239] It should be noted that the above examples are only to better illustrate the solution of the present invention, rather than to limit the present invention. The real popularity can also be characterized by other ways, for example, using ranks for characterization, etc., any according to the input sequence The implementation of obtaining candidate search result items and their true popularity corresponding to the input sequence should all be included in the scope of the present invention.

[0240] The sorting device 3 sorts the candidate search result items according to the true popularity to obtain the sorting result.

[0241] Specifically, the sorting device 3 sorts the candidate search result items corresponding to the real popularity in a descending manner of the real popularity. Among them, the way from high to low includes but is not limited to:

[0242] 1) When the true popularity is represented by a value, the candidate search result items are sorted according to the value of the true popularity from high to low;

[0243] 2) When the real popularity is represented by a rank, the candidate search result items are sorted according to the rank of the real popularity from high to low.

[0244] It should be noted that the above examples are only to better illustrate the solution of the present invention, rather than to limit the present invention. Any sorting of the candidate search result items according to the true popularity to obtain the realization of the ranking results All methods should be included in the scope of the present invention.

[0245] The method according to the present invention can be used to feed back the sorting result according to the user's input sequence in real-time online, and can also be used for offline processing to obtain the sorting result corresponding to the input sequence. Further, the method according to the present invention can be used to feed back the sorting results according to the query input sequence entered by the user in the search box, for example, by inputting the query input sequence in the search box provided by the search engine or website to obtain the result via the Internet Sort the results, or perform resource query through web pages or clients in the local area network; or, in the input method, provide users with sorted short texts according to the input sequence input by the user.

[0246] Since the method according to the present invention refers to the real popularity to determine the sorting result, it can better reflect the real concerns of the user compared with the prior art, so that the sorting result is more in line with the user's needs.

[0247] As one of the preferred solutions of the present invention, the present invention also includes a determining device (not shown in the figure). The determining device determines the true popularity corresponding to the candidate search result item under the input sequence according to at least one of the following:

[0248] 1) The relative click rate of the candidate search result item corresponding to the input sequence;

[0249] Specifically, the determining device obtains the relative click-through rate according to the recorded number of clicks and display times of the candidate search result item corresponding to the input sequence.

[0250] For example, the determining device is based on the formula:

[0251] Relative click rate=k*(number of clicks/number of impressions) to obtain relative click rate;

[0252] Among them, k is a parameter that can be determined by a person skilled in the art according to actual conditions and requirements.

[0253] For another example, the determining device is based on the formula:

[0254] Relative click rate = clicks/(number of impressions-clicks) to obtain the relative click rate.

[0255] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any item corresponding to the input sequence according to the candidate search result The click-through rate to obtain the real popularity should be included in the scope of the present invention.

[0256] Wherein, the number of clicks indicates the number of times the user clicks on the candidate search result item under the input sequence, and the number of presentations indicates the number of times the candidate search result item is presented to the user under the input sequence.

[0257] Specifically, the search device or other device obtains the number of times of display by recording the number of times the candidate search result item is displayed in front of the user under each input sequence. When the number of presentations needs to be processed, the determining device queries the recorded corresponding number of presentations according to the input sequence and candidate search result items.

[0258] Among them, the ways to judge the candidate search result items to be displayed in front of the user include but are not limited to:

[0259] i) Determine that the candidate search result item presented through the browser is the candidate search result item displayed in front of the user;

[0260] The search device or other devices provide the browser with data for displaying to the user directly or via the user equipment according to the user's confirmation of the search behavior and the input sequence input by the user. Then, the search device or other devices determine the candidate search result items included in the data provided to the browser as candidate search result items displayed in front of the user under the input sequence.

[0261] For example, the user accesses the Baidu search engine through the user device, enters the input sequence "game download" and clicks the button "Baidu Click" to confirm the search. Then, the Baidu search engine determines that the candidate search result items included in the data fed back to the user equipment according to the input sequence "game download" are candidate search result items displayed in front of the user under the input sequence "game download".

[0262] ii) Determine that the candidate search result item presented by the client is the candidate search result item presented in front of the user;

[0263] The search device or other device directly or via the user equipment provides the client with data for presenting to the user according to the user's behavior. Then the search device or other devices determine the candidate search result items included in the data provided to the client as candidate search result items displayed in front of the user.

[0264] For example, a user accesses a local search engine of a paper library through a paper search client, enters the input sequence "game download" and clicks the button "confirm" to confirm the search. Then the local search engine judges that the candidate search result items contained in the data fed back to the client according to the input sequence "game download" are candidate search result items displayed in front of the user under the input sequence "game download".

[0265] iii) Determine that the displayed input method candidates are candidate search result items displayed in front of the user;

[0266] When the candidate search result item is an input method candidate, the search device or other device judges the input method candidate included in the data used to generate the input method candidate column as the candidate search result displayed in front of the user under the input sequence item.

[0267] For example, the input sequence input by the user is "youxi", and the input method candidates included in the data used to generate the input method candidate column obtained by the user device based on the input sequence are "game", "youxi", and "youxi" , "Youxi" and "youxi", the search device or other device determines that the above five input method options are the candidate search result items displayed in front of the user under the input sequence "youxi". It should be noted that when the user performs the operation of presenting the next page, for example, selects the "pagedown" button, etc., the input method candidates included in the data in the update input method candidate column are also judged as being in the input sequence. Candidate search result items displayed in front of users under "youxi".

[0268] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any number of candidate search result items displayed in front of the user by recording each input sequence All the ways to achieve the number of times of presentation should be included in the scope of the present invention.

[0269] The search device or other devices also obtain the number of clicks by the user by recording the number of times each candidate search result item under each input sequence is clicked. When the number of clicks needs to be processed, the determining device queries the recorded corresponding number of clicks according to the input sequence and candidate search result items. The search device or other device uses the number of times selected by the user as the number of clicks during the process of displaying the candidate search result items, for example, during the process of displaying the candidate search result items through a browser, client or input method candidate column. Those skilled in the art should be able to determine the implementation manner of obtaining the number of clicks, which will not be repeated here.

[0270] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any click corresponding to the input sequence according to the candidate search result item The number of times and the number of times of presentation to achieve the relative click rate should be included in the scope of the present invention.

[0271] 2) The number of user recommendations corresponding to the candidate search result item and the input sequence;

[0272] Specifically, when a device providing candidate search result items provides a function of recommending each candidate search result item to tourists or registered users, the device records the number of recommendations for each candidate search result item obtained based on the input sequence and recommended by the user. When the number of recommendations needs to be processed, the determining device queries the number of recommendations recorded according to the input sequence and candidate search result items, and obtains the true popularity according to the number of recommendations, for example, directly regard the number of recommendations as the true Popularity, or after processing the recommended number, for example, normalizing or dividing by a predetermined coefficient, the processed value is used as the true popularity.

[0273] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any user who corresponds to the input sequence according to the candidate search result item The number of recommendations to obtain the real popularity should be included in the scope of the present invention.

[0274] 3) The number of user favorites corresponding to the candidate search result item and the input sequence;

[0275] Specifically, when the device that provides candidate search result items provides a function for registered users to bookmark each candidate search result item, the device records the number of favorite search result items obtained by the user based on the input sequence. When the number of favorites needs to be processed, the determining device queries the recorded number of favorites according to the input sequence and candidate search result items, and obtains the true popularity according to the number of favorites, for example, directly regards the number of favorites as true Popularity, or after processing the number of favorites, for example, normalizing or dividing by a predetermined coefficient, the processed value is used as the true popularity.

[0276] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any user who corresponds to the input sequence according to the candidate search result item The number of favorites to obtain the real popularity should be included in the scope of the present invention.

[0277] It should be noted that, the determining device may also combine any number of the relative click rate, the number of user recommendations, and the number of user favorites to obtain the true popularity. For example, the search device normalizes the relative click-through rate, the number of user recommendations, and the number of user favorites respectively, and then obtains the average value or sum of the three above or performs corresponding processing according to a preset formula to obtain the true popularity. Those skilled in the art should understand that the scope of the present invention is not limited to the foregoing examples.

[0278] It should be further explained that the way in which the determining device determines the true popularity can also be used for the sorting device to obtain the pre-stored multiple factors related to the true popularity and then obtain the true popularity according to the multiple factors.

[0279] As one of the preferred solutions of the present invention, the sorting device 3 also sorts the candidate search result items based on the first predetermined rule and according to the true popularity to obtain the sorting result.

[0280] Specifically, the first predetermined rule includes a sorting rule for determining the sorting result of the candidate search result item based on one or more existing factors in combination with the true popularity. Then, the sorting device 3 obtains the one or more existing factors according to the sorting rule specified by the first predetermined rule, and determines the candidate search result item according to the obtained existing factors in combination with the true popularity. The sorted results.

[0281] Among them, the existing factors include but are not limited to:

[0282] 1) The degree of correlation between the input sequence and the candidate search result item;

[0283] The correlation degree may be obtained according to factors such as the degree of matching between the input sequence and the candidate search result item, the category of the input sequence, and/or the category of the candidate search result item;

[0284] 2) The importance of the candidate search result item;

[0285] The importance can be obtained according to factors such as click-through rate and authority of candidate search result items.

[0286] Since the above two factors are all existing factors, those skilled in the art should be able to obtain the above two factors based on the input sequence and the candidate search result items according to actual conditions and needs. Therefore, the two factors are not mentioned here. The method of obtaining the factors is described in detail.

[0287] It should be noted that the above-mentioned examples of existing factors are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any other existing factors that affect the ranking of candidate search results All factors should be included in the scope of the existing factors described in the present invention.

[0288] The sorting rules further include any of the following rules:

[0289] 1) First determine the initial ranking result of the candidate search result items according to existing factors, and then adjust the initial ranking result according to the true popularity to obtain the rules of the ranking result;

[0290] For example, for the input sequence "game download", the initial ranking result of the candidate search result items obtained by the ranking device 3 according to existing factors is:

[0291] Candidate search result item one;

[0292] Candidate search result item three;

[0293] Candidate search result item four;

[0294] Candidate search result item two;

[0295] Candidate search result item five.

[0296] The true popularity of each candidate search result item obtained by the second obtaining device 2 and the input sequence "game download" is as follows:

[0297] Candidate search result item True popularity

[0298] Candidate search result item one first level

[0299] Candidate search result item 2 Level 1

[0300] Candidate search result item 3 Level 3

[0301] Candidate search result item 4 Second level

[0302] Candidate search result item five third level

[0303] If the sorting rule includes analyzing the true popularity of two adjacent candidate search result items from front to back, when the true popularity level of the candidate search result item is higher than that of the previous candidate search result item, the two positions will be mutually Change; when the true popularity level of the candidate search result item in the back is lower than or equal to the candidate search result item in the front, the rules of the two ranking positions are not changed.

[0304] According to the above rules, the ranking device 3 analyzes that the real popularity level of the candidate search result item 1 is higher than the real popularity ranking of the candidate search result item 3, and the ranking positions of the two are not changed; then, the ranking device 3 analyzes and obtains the candidate search results The true popularity level of item three is lower than the true popularity level of candidate search result item four, the positions of candidate search result item three and candidate search result item four are swapped; then, because the position of candidate search result item three has changed The original second place is changed to the third place. Therefore, the sorting device 3 compares the true popularity level of the candidate search result item three with the true popularity level of the candidate search result item 2, and judges the authenticity of the candidate search result item 3. If the popularity level is lower than the true popularity level of candidate search result item 2, the positions of candidate search result item 3 and candidate search result item 2 are swapped; then, because the position of candidate search result item 3 has been changed from third To the fourth place, therefore, the sorting device 3 compares the real popularity level of candidate search result item 3 with the real popularity level of candidate search result 5, and determines that the real popularity level of candidate search result item 3 is equal to the candidate search The actual popularity level of the result item 5 does not change the ranking position of the two. After the sorting device 3 performs the above operations, the result of sorting candidate search terms is:

[0305] Candidate search result item one;

[0306] Candidate search result item four;

[0307] Candidate search result item two;

[0308] Candidate search result item three;

[0309] Candidate search result item five.

[0310] 2) Synthesize the existing factors and the real popularity to directly obtain the rules of the ranking result.

[0311] For another example, for the input sequence "game download", the existing factors referred to by the sorting rule include the relevance of the input sequence and the candidate search result item and the importance of the candidate search result item, and the ranking The rule stipulates that the relevance, importance, and true popularity are added, and the candidate search result items are sorted according to the sum obtained. If the relevance and importance obtained by the ranking device 3 and the real popularity obtained by the second obtaining device 2 are as follows:

[0312] Candidate search results Item relevance Importance True popularity

[0313] Candidate search result item one 2.135 2.167 1.234

[0314] Candidate search result item two 3.214 3.246 1.456

[0315] Candidate search result item three 2.489 2.498 2.487

[0316] Candidate search result item four 1.358 1.397 1.112

[0317] Candidate search result item five 1.698 2.054 1.587

[0318] Then the sorting device 3 calculates the sum of the relevance, importance, and true popularity of each candidate search result item according to the rules of the sorting rule, and sorts the candidate search result items according to the added sum, and the sorting result obtained is:

[0319] Candidate search result item two;

[0320] Candidate search result item three;

[0321] Candidate search result item one;

[0322] Candidate search result item five;

[0323] Candidate search result item four.

[0324] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any method based on the first predetermined rule and according to the actual popularity will be The candidate search result items are sorted to obtain the realization method of the sorting result, for example, the rule of adjusting the initial sorting result according to the real popularity is changed, or, when the real popularity is represented in the form of value, according to the real popularity To adjust the initial ranking result, or to weight each existing factor and real popularity, and then multiply or average the weighted existing factors and real popularity to obtain the All sorting values ​​should be included in the scope of the present invention.

[0325] As one of the preferred solutions of the present invention, the second obtaining device further includes an information unit obtaining device (not shown) and a search result obtaining device (not shown).

[0326] The information unit obtaining device obtains the information units included in the input sequence.

[0327] Specifically, the manner in which the information unit obtaining apparatus obtains the information unit includes but is not limited to:

[0328] 1) The information unit obtaining device segments the input sequence to obtain a word segmentation unit as an information unit;

[0329] 2) The information unit obtaining device extracts the n-gram segment of the input sequence to use the n-gram segment as the information unit.

[0330] Further, the information unit obtaining device can also remove invalid information in the word segmentation unit or n-gram segment obtained by the above two methods, so as to use the remaining word segmentation unit or n-gram segment as the information unit.

[0331] The search result obtaining device obtains the candidate search result item according to the information unit.

[0332] Specifically, the search result obtaining device first obtains candidate search result items according to the information unit, and then uses the candidate search result items obtained based on the information unit as the candidate search result items of the input sequence to which the information unit belongs. The manner in which the search result obtaining device obtains candidate search result items according to the information unit is the same as or similar to the manner in which the second obtaining device 2 obtains candidate search result items according to the input sequence, and will not be repeated here.

[0333] As one of the preferred solutions of the present invention, the search device further includes an update device (not shown), a real-time acquisition device (not shown), and a query device (not shown).

[0334] The updating device establishes or updates the ranking index information according to the input sequence and the ranking result of the corresponding candidate search result items.

[0335] Wherein, the sort index information includes any one of the following index information:

[0336] 1) The input sequence, the candidate search result item corresponding to the input sequence, and the ranking position of the candidate search result item under the input sequence;

[0337] Specifically, the updating device directly establishes or updates the input sequence, the candidate search result item corresponding to the input sequence, and the index information of the ranking position of the candidate search result item under the input sequence according to the ranking result obtained by the ranking device 3.

[0338] 2) The input sequence, the candidate search result item corresponding to the input sequence, and the ranking position adjustment information of the candidate search result item under the input sequence;

[0339] Specifically, the updating device obtains the ranking position adjustment information of the candidate search result item under the input sequence according to the real popularity or the rank or weight value of other existing factors contained in the ranking result obtained by the ranking device 3, for example, weighting Information or weight reduction information, etc., to establish or update the input sequence, the candidate search result item corresponding to the input sequence, and the index information of the position adjustment information of the candidate search result item sorted under the input sequence.

[0340] The real-time acquisition device acquires the input sequence from the user in real time.

[0341] The query device performs a matching query in the sort index information according to the input sequence obtained in real time to obtain a sort result corresponding to the input sequence obtained in real time.

[0342] Specifically, the query device directly queries the input sequence to obtain the ranking result of the input sequence based on the input sequence contained in the ranking index information, the candidate search result item corresponding to the input sequence, and the ranking position of the candidate search result item under the input sequence. .

[0343] Alternatively, the query device processes the candidate search result items according to the input sequence contained in the ranking index information, the candidate search result item corresponding to the input sequence, and the ranking position adjustment information of the candidate search result item under the input sequence to Get the sort result of the input sequence.

[0344] Preferably, the present invention further includes a selection device, which selects a search result item to be presented to the user from the candidate search result items according to the ranking result.

[0345] Specifically, the selection device selects the search result items to be presented to the user from the candidate search result items according to the user behavior and the number of candidate search result items that need to be presented.

[0346] Figure 5 It is a schematic structural diagram of a search device for improving search results based on user behavior in a preferred embodiment of the present invention. The search device according to this embodiment includes a first acquisition device 1, a second acquisition device 2, and a first sub-sorting device 31 included in the sorting device 3.

[0347] The first acquisition device 1 and the second acquisition device 2 are already referencing Figure 4 The illustrated embodiment is described in detail, and is included here by reference, and will not be repeated.

[0348] The first sub-sorting device 31 sorts the candidate search result items based on the first predetermined rule and according to the real popularity and other sorting related information to obtain the sorting result.

[0349] The other sorting related information includes at least one of the following:

[0350] 1) Other user behavior information related to the candidate search result item;

[0351] The other user behavior information can be obtained by a search device or other devices, and the first sub-sorting device 31 can obtain the other user behavior information obtained by other devices through a communication connection or other connection methods. The other user behavior information includes the following At least one:

[0352] i) The presentation time of the candidate search result item corresponding to the input sequence;

[0353] The presentation time indicates the time that the candidate search result item is displayed in front of the user. Specifically, when the candidate search result item is displayed through the browser, the display time represents the time from when the browser generates the page to when the user updates the candidate search result item on the page or closes the page; when the candidate search result item passes the client When the client is displayed, the display time represents the time from when the client generates the display page to when the user updates the candidate search result item in the display page or closes the page; when the candidate search result item is displayed through the input method candidate column, the display time Represents the time when the candidate search result item is displayed in the input method candidate column.

[0354] ii) The display position of the candidate search result item on the page;

[0355] The display position includes the position of the candidate search result item in the web page, client page, or input method candidate column when it is displayed to the user. For example, for web pages and client pages, the display position includes upper, left, right, lower, middle, title, text, sidebar, etc. For input method candidate columns, the display position includes input How many options are in the candidate column?

[0356] iii) The browsing time corresponding to the resource corresponding to the candidate search result item and the input sequence;

[0357] The browsing time indicates the length of time for the user to browse the candidate search result item after selecting it. For example, for a browser or client to present a candidate search result item, the browsing time can be based on the time for browsing the resource after the user opens the resource corresponding to the candidate search result item through a click or keyboard selection behavior; where The time when the resource is displayed is used to obtain the browsing time. For example, if the resource is a webpage, the obtained time of existence of the webpage is used as the user's browsing time.

[0358] Specifically, the first sub-sorting device 31 adjusts the sorting result based on the first predetermined rule according to at least one item of other user behavior information and based on the real popularity; or, the first sub-sorting device 31 directly A predetermined rule refers to existing factors, real popularity, and other user behavior information to obtain ranking results.

[0359] Among them, the longer the candidate search result item is displayed, the higher the ranking; the farther the candidate search result item is displayed, and the farther it is selected by the user, the higher the ranking. It is assumed that, for example, the middle part and title of the preset page, the first item in the input method candidate column is the user's attention area, etc.; the longer the browsing time of the candidate search result item, the higher the ranking. The first sub-sorting device 31 can determine the adjustment level to which the display time, the display position, and the browse time belong according to any one or several items of the obtained information of the display time, the display position, and the browsing time, and according to The adjustment level is used to adjust the foregoing sorting result, and the way of adjusting the sorting result according to the adjustment level is in accordance with the reference figure 1 In the illustrated embodiment, the method of adjusting the initial ranking results according to the actual popularity is the same or similar, and is included here by reference, and will not be repeated; or, the search device is based on the obtained display time, display position, and browsing time. Any item or any number of items in the information is used to determine the individual weight value corresponding to the display time, display position, and browsing time, and to obtain the sorting result according to the individual weight value. This method will be described in detail in subsequent embodiments. I will not repeat them here.

[0360] 2) User related information;

[0361] Among them, the user's identity can be identified according to the user's registration information; the identification code of the user's device, such as mobile phone number or hardware hash value, etc.; and the user's temporary record information, such as user cookie. The user-related information includes at least one of the following:

[0362] i) Personal attributes of the user;

[0363] The personal attributes include, but are not limited to, the age, gender, occupation, income, education level and other information of the user actively provided by the registered user or inferred from the recorded user behavior. The first sub-sorting device 31 adjusts the sorting result according to the user's personal attributes. For example, if the user is a female, the candidate search result items whose category is "shopping" are sorted in advance. Among them, the category of the candidate search result item is obtained through preset information.

[0364] ii) User preferences;

[0365] The first sub-sorting device 31 adjusts the sorting result according to the user's preference setting. For example, if the preference setting includes a setting to advance candidate search result items of the "military" category. Then, the first sub-sorting device 31 advances the sorting position of the candidate search result items of the "military" category in the sorting result according to the preference setting.

[0366] iii) User's environmental information;

[0367] The environmental information includes time information and location information. Wherein, the time information may be obtained through the current time of the user equipment, and the location information may be obtained according to the current location information of the user equipment, for example, IP address or mobile phone roaming area. The first sub-sorting device 31 adjusts the sorting result according to the environmental information. For example, if the first sub-sorting device 31 analyzes and finds that the current IP address of the user is Shanghai, and the user input sequence is "weather forecast", then the weather forecast sorting position in the Shanghai area is advanced, etc.

[0368] iv) User's historical behavior record;

[0369] The user's historical behavior record includes but is not limited to: user's click behavior, user's browsing behavior, etc. The first sub-sorting device 31 adjusts the sorting result according to the user's historical behavior record. For example, if the historical behavior record includes the record of the user selecting the weather forecast in the Shanghai area when inputting the sequence "weather forecast", when the user's input sequence is obtained as "weather forecast", the weather forecast in the Shanghai area will be sorted. Wait in advance.

[0370] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any based on the first predetermined rule, according to the real popularity and other ranking related Information to sort the candidate search result items to obtain the sorting result should all be included in the scope of the present invention.

[0371] Image 6 It is a schematic structural diagram of a search device for improving search results based on user behavior in another preferred embodiment of the present invention. The search device according to this embodiment includes a first acquiring device 1, a second acquiring device 2, and a sorting device 3. The ranking device 3 further includes a weight obtaining device 32 and a second sub-ranking device 33.

[0372] The first acquisition device 1 and the second acquisition device 2 are already referencing Figure 4 It is described in detail in the illustrated embodiment, and is included here by reference, and will not be repeated.

[0373] In this embodiment, the first predetermined rule further includes a weight ordering rule.

[0374] The weight obtaining device 32 obtains the weight value of the candidate search result item under the input sequence based on the weight sorting rule, according to the real popularity and the other sorting related information.

[0375] Specifically, the weight ranking rule refers to existing factors, real popularity, and other related ranking information to determine the weight value. Among them, the existing factors, real popularity and other ranking related information have been referenced Figure 4 or Figure 5 It is described in detail in the illustrated embodiment, and is included here by reference, and will not be repeated. The weight ordering rule further includes any one of the following rules:

[0376] 1) The individual weight value corresponding to any one of the existing factors, real popularity or other ranking related information is used as the initial weight value, and then the initial weight value is adjusted according to the individual weight values ​​corresponding to the other two to obtain the aforementioned Weights.

[0377] For example, according to the weight ranking rule, the weight obtaining device 32 uses the individual weight value corresponding to the existing factor as the initial weight value, and then, according to the actual popularity and other ranking related information, the individual weight value corresponds to the individual weight value corresponding to the existing factor The predetermined range to which the difference value of the value belongs is used to determine the upward or downward adjustment range of the initial weight value. Among them, those skilled in the art should be able to determine the predetermined range according to actual conditions and requirements. For example, if the individual weight values ​​corresponding to the true popularity and other ranking related information are greater than the individual weight values ​​corresponding to the existing factors, and the difference falls within a predetermined range between 0.2 and 0.4, the initial weight value is increased by 0.1, etc.

[0378] Among them, those skilled in the art should be able to determine the individual weight value of the existing factor according to the actual situation, which will not be repeated here.

[0379] The individual weight value corresponding to the true popularity is based on the relative click rate of the candidate search result item and the input sequence, the number of user recommendations corresponding to the candidate search result item and the input sequence, and the candidate search result The item is obtained from at least one of the user favorites corresponding to the input sequence. For example, the weight acquiring device 32 directly uses the value corresponding to the relative click rate, the number of recommendations, or the number of favorites as the weight value of the real popularity; for another example, the weight acquiring device 32 normalizes the relative click rate, the number of recommendations, and the number of favorites, respectively After addition, the sum obtained is used as the weight value corresponding to the real popularity. Those skilled in the art should understand that the method of obtaining the individual weight value corresponding to the real popularity is not limited to the above example.

[0380] The individual weight values ​​corresponding to other ranking related information are obtained according to one or more types of information contained in the other ranking related information. For example, the weight obtaining device 32 directly uses any one of the values ​​corresponding to each item in other user behavior information as the weight value of other ranking related information; for another example, the weight obtaining device 32 separates the values ​​corresponding to each item of other user behavior information. Perform normalization, and obtain the average value of each normalized value as the individual weight value corresponding to other ranking related information. Those skilled in the art should understand that the method for obtaining the individual weight values ​​corresponding to other sorting related information is not limited to the above examples.

[0381] 2) Obtain the aforementioned weight value directly according to the individual weight value corresponding to at least one of the existing factors, real popularity or other ranking related information.

[0382] For example, the weight ranking rules stipulate that the weighting coefficients of existing factors, real popularity or other ranking-related information are 0.8, 1.2, and 1.1 respectively, and stipulate that the weight values ​​of each individual item are multiplied by the corresponding weighting coefficients to obtain the weight value. Then, the search device multiplies the weight values ​​of the individual items by the corresponding weight coefficients and adds them, and uses the sum obtained by the addition as the weight value.

[0383] It should be noted that the above examples are only to better illustrate the technical solutions of the present invention, rather than to limit the present invention. Those skilled in the art should understand that any sorting rules based on the weights are based on the true popularity and the Other sorting-related information to obtain the weight value of the candidate search result item under the input sequence, such as using other weight adjustment methods or other weight calculation methods, should be included in the scope of the present invention .

[0384] The second sub-sorting device 33 sorts the candidate search result items according to the weight value to obtain the sorting result.

[0385] Specifically, the second sub-sorting device 33 sorts the candidate search result items in a descending order of the weight value to obtain the sorting result.

[0386] For those skilled in the art, it is obvious that the present invention is not limited to the details of the foregoing exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the foregoing description, and therefore it is intended to fall within the claims. All changes within the meaning and scope of the equivalent elements of are included in the present invention. Any reference signs in the claims should not be regarded as limiting the claims involved. In addition, it is obvious that the word "including" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claims can also be implemented by one unit or device through software or hardware. Words such as first and second are used to denote names, but do not denote any specific order.