An information publishing service system and method

Abstract

The present invention discloses an information publishing service system and method, which relates to the technical field of information publishing. The present invention obtains the user's historical information browsing records, divides the historical information browsing records according to different ways of obtaining information, and obtains a set of historical information browsing records in any way; obtains the reading parameters after the user browses the information, calculates the reading index of the information, and determines that the information is characteristic information; confirms the type of the characteristic information, calculates the weight of any entry of any information type, and divides the level of the entry; the user's score for the information and the number of entries at any level form a data group, and the data group is subjected to function fitting to obtain a score prediction model for any information type; according to the score prediction model, the information to be published in real time is sorted to achieve precise push of information.

Description

Technical Field

[0001] The present invention relates to the technical field of information publishing, and particularly to an information publishing service system and method. Background Art

[0002] Nowadays, we are in the era of informatization and dataization. With the continuous deepening of social informatization, the increase in user types and the convenience of obtaining information services have greatly stimulated the potential of users' information needs and the requirements for diversified information services.

[0003] Information publishing platforms have been integrated into life. On information publishing platforms, the latest news will be released to users. However, different users have different interests, so there will be certain demand differences. If the user's needs are not considered when publishing information, it will lead to a poor reading experience for users when reading information. Therefore, it is particularly important to develop a publishing service system that can combine the personal preferences of users, automatically match the information to be published with user information, and achieve precise push. Summary of the Invention

[0004] The purpose of the present invention is to provide an information publishing service system and method to solve the problems raised in the prior art.

[0005] To solve the above technical problems, the present invention provides the following technical solution: An information publishing service method, the method includes:

[0006] Step S100: Through the information publishing platform system, obtain the user's historical information browsing records, collect the information acquisition methods of the historical information browsing records, divide the historical information browsing records according to different information acquisition methods, and obtain a set of historical information browsing records in any way.

[0007] Step S200: In the set of historical information browsing records in any way, obtain the reading parameters after the user browses the information, calculate the reading index of the information through calculation, and determine whether the information is characteristic information.

[0008] Step S300: According to the information type of the information publishing platform system, confirm the type of the characteristic information, perform text processing on the characteristic information of any type, obtain the composition of the entries of the characteristic information and the proportion of the entries, and calculate the weight of any entry.

[0009] Step S400: According to the weight of any entry in step S300, calculate the weight of any entry of any information type, and divide the level of the entry according to the weight.

[0010] Step S500: Obtain the rating records of the user for a certain piece of historical browsing information. In the rating records, form a data group by combining the user's rating of the information and the number of entries of any level, and perform function fitting on the data group to obtain a rating prediction model for any type of information.

[0011] Step S600: Sort the information to be released in real time according to the rating prediction model, and push the information to the user in the arranged order.

[0012] Further, in step S100, in the information release platform system, obtain the user's historical information browsing records, and summarize the historical information browsing records obtained in the same way of obtaining information to obtain a set A of historical information browsing records in any way = {a1, a2, a3,..., an}, where a1, a2, a3,..., an respectively represent the 1st, 2nd, 3rd,..., nth historical information browsing records in any way; among them, an = {an1, an2, an3,..., anm}, where an1, an2, an3,..., anm respectively represent the numerical values of the 1st, 2nd, 3rd,..., mth reading parameters in the nth historical information browsing record in any way.

[0013] When the user browses information, the ways of obtaining information are different, including active search by the user, system automatic recommendation, etc. Different ways may represent different degrees of user interest, so it is necessary to conduct classified discussions to improve the accuracy of data for subsequent calculations.

[0014] The reading parameters in the information browsing record include reading duration, browsing times, reading depth, etc. Among them, the reading depth is the depth of the user's scrolling page, so as to judge whether the user has browsed the entire article or only read the beginning part.

[0015] Further, step S200 includes:

[0016] Step S201: Calculate the reading index of the nth historical information browsing record in the set of historical information browsing records in any way according to the formula:

[0017]

[0018] where, X bn represents the reading index of the nth historical information browsing record in the bth way, X bnc represents the numerical value of the cth reading parameter in the nth historical information browsing record in the bth way, and d bc represents the weight of the cth reading parameter in the bth way.

[0019] Step S202: When X bn ≥X bWhen X b represents the reading index threshold in the b-th mode, it is determined that the n-th historical information browsing record in the b-th mode is the characteristic information in the b-th mode. The characteristic information of any mode is summarized to obtain the characteristic information set of any mode and the reading index set of the characteristic information;

[0020] Because the importance of different reading parameters is different, it is necessary to calculate the reading index of the information browsing record, determine whether the information is characteristic information, and analyze the characteristic information to ensure the accuracy of subsequent analysis.

[0021] Further, step S300 includes:

[0022] Step S301: According to the information type of the information release platform, collect the information types of the characteristic information in the characteristic information set of any mode, and summarize a certain information type obtained by collection to obtain the characteristic information set of any information type;

[0023] Step S302: Input the characteristic information in the characteristic information set of any information type into the NLP model for text analysis, extract the entries of the characteristic information, summarize the number of times any entry appears in the characteristic information set, and calculate the frequency of a certain entry in any characteristic information according to the following formula:

[0024]

[0025] Where Y tm represents the frequency of the m-th entry in the t-th characteristic information, y tm represents the number of times the m-th entry appears in the t-th characteristic information, and y t represents the total number of all entries in the t-th characteristic information;

[0026] Step S303: In the reading index set of the characteristic information, calculate the weight of a certain entry in any characteristic information according to the following formula:

[0027] Z tm =Y tm ×X t ;

[0028] Where Z tm represents the weight of the m-th entry in the t-th characteristic information, and X t represents the reading index of the t-th characteristic information, obtaining the entry weight set of any characteristic information, and summarizing the characteristic information of any information type to obtain the weight set of a certain entry in any information type;

[0029] The information types include people's livelihood, sports, education, etc.;

[0030] Since the frequency of a term only represents part of the data, if only the frequency of the term is analyzed, the result will be single and the accuracy will decrease. Therefore, it is necessary to analyze it together with the reading index, and the result obtained in this way will be more accurate.

[0031] Furthermore, step S400 includes:

[0032] Step S401: Divide the weight set of a certain term of the same information type according to any way of obtaining information, and obtain the weight set of a certain term in any way in the same information type;

[0033] Step S402: Calculate the weight of a certain term in any information type according to the following formula:

[0034]

[0035] where, Z pm represents the weight of the m-th term in the p-th information type, and Z rpmi represents the i-th weight of the m-th term in the p-th information type in the r-th way, and j rpm represents the number of weight sets of the m-th term in the p-th information type in the r-th way, and α r represents the weight of the r-th way, and L represents the number of ways of obtaining information, and the weight set of terms in any information type is obtained;

[0036] Step S403: Divide the term weights into several levels, divide the weight set of terms in any information type according to the levels, confirm the term levels in any information type and summarize them to obtain a set of terms at a certain level in any information type;

[0037] Since the information obtained by different methods represents different degrees of user interest, the weights of different methods need to be calculated together to obtain more accurate weights.

[0038] Furthermore, step S500 includes:

[0039] Step S501: Obtain the scoring record of the user for a certain historical browsing information, collect the information type and terms of the browsing information, determine the set of terms at any level according to the collected information type, compare the collected terms with the set of terms at any level, and obtain the number of terms at any level;

[0040] Step S502: Take the score of the information and the number of terms at any level as a data group, summarize the scoring records of the historical browsing information, and perform function fitting on several data groups according to the following formula:

[0041]

[0042] Among them, H represents the score of the information, and S e represents the number of entries at the e-th level, and β e represents the weight at the e-th level, f is the number of levels, S represents the total number of entries that meet any level in any information, S1 represents the total number of entries in any information, and γ represents the weight of the information score, thus obtaining the prediction score model for any information type.

[0043] Furthermore, step S600 includes:

[0044] Step S601: Obtain the information type and entries of the real-time information to be released, determine the number of entries that meet any level for a certain user, input the number of entries at the any level into the prediction score model of the same information type, and obtain the predicted score of the user;

[0045] Step S602: When the predicted score of the user is greater than or equal to the score threshold, set the information to be released as the interested information, obtain the set of interested information, sort the set according to the predicted scores of the interested information from high to low, and push the sorted set of interested information to the user.

[0046] In order to better implement the above method, an information release service system is also proposed. The system includes a historical information browsing record module, a feature information module, a term weight module, a term level module, a score prediction model module, and a real-time sorting module;

[0047] Historical information browsing record module: Through the information release platform system, obtain the user's historical information browsing records, collect the information acquisition methods of the historical information browsing records, divide the historical information browsing records according to different information acquisition methods, and obtain the set of historical information browsing records in any method;

[0048] Feature information module: In the set of historical information browsing records in any method, obtain the reading parameters of the user after browsing the information, calculate the reading index of the information through calculation, and determine whether the information is feature information;

[0049] Term weight module: According to the information type of the information release platform system, confirm the type of the feature information, perform text processing on the feature information of any type, obtain the term composition and term proportion of the feature information, and calculate the weight of any term;

[0050] Term level module: According to the weight of any term in step S300, calculate the weight of any term in any information type, and divide the level of the term according to the weight;

[0051] Rating prediction model module: Obtain the rating records of a user for a certain historical browsing information. In the rating records, form a data group by combining the rating of the user for the information and the number of entries of any level, and perform function fitting on the data group to obtain a rating prediction model for any information type;

[0052] Real-time sorting module: Sort the information to be released in real time according to the rating prediction model, and push the information to the user in the arranged order.

[0053] Furthermore, the entry weight module includes a feature information classification unit and an entry weight calculation unit;

[0054] Feature information classification unit: Collect the information types of feature information according to the information types of the information release platform, summarize a certain information type obtained by the collection, and obtain a set of feature information for any information type;

[0055] Entry weight calculation unit: Extract entries from the feature information through a natural language processing model, calculate the frequency of the entry appearing in the feature information, and calculate the weight of the entry according to the frequency and the reading index of the feature information.

[0056] Furthermore, the entry level module includes an entry weight calculation unit and a level determination unit;

[0057] Entry weight calculation unit: Obtain the set of weights of a certain entry in any way in the same information type, calculate the set of weights, and obtain the weight of a certain entry in any information type;

[0058] Level determination unit: Divide the entry weights into several levels, divide the set of entry weights of any information type according to the levels, confirm and summarize the entry levels of any information type, and obtain a set of entries of a certain level in any information type.

[0059] Compared with the prior art, the beneficial effects of the present invention are: Through the information release platform system, obtain the user's historical information browsing records, collect the information acquisition methods of the historical information browsing records, divide the historical information browsing records according to different information acquisition methods, obtain a set of historical information browsing records in any way, obtain the reading parameters of the user after browsing the information, and calculate the reading index of the information, and determine whether the information is feature information. This method for determining feature information first avoids subjective human judgment, and secondly combines the reading parameters and the entry frequency, improving the accuracy of the determination;

[0060] According to the information type of the information release platform system, confirm the type of the feature information, perform text processing on any type of feature information, obtain the composition of the entries of the feature information and the proportion of the entries, calculate the weight of any entry, calculate the weight of any entry of any information type according to the weight of any entry in step S300, and divide the level of the entry according to the weight. By using this method to divide the entry level, the accuracy of the division is further improved;

[0061] Obtain the scoring records of a user for a certain historical browsing information. In the scoring records, form a data group with the user's score for the information and the number of entries of any level, perform function fitting on the data group to obtain a scoring prediction model for any information type; sort the information to be released in real time according to the scoring prediction model. Through this method, accurate real-time information push can be achieved. Brief Description of the Drawings

[0062] Figure 1 It is a schematic flowchart of a method for an information release service according to the present invention. Detailed Embodiment

[0063] Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0064] Please refer to Figure 1 , the present invention provides a technical solution: a method for an information release service, the method includes:

[0065] Step S100: Through the information release platform system, obtain the user's historical information browsing records, collect the information acquisition methods of the historical information browsing records, and divide the historical information browsing records according to different information acquisition methods to obtain a set of historical information browsing records in any way;

[0066] Among them, in step S100, in the information release platform system, obtain the user's historical information browsing records, and summarize the historical information browsing records of the same information acquisition method to obtain a set of historical information browsing records in any way A = {a1, a2, a3,..., an}, where a1, a2, a3,..., an respectively represent the 1st, 2nd, 3rd,..., nth historical information browsing records in any way; among them, an = {an1, an2, an3,..., anm}, where an1, an2, an3,..., anm respectively represent the values of the 1st, 2nd, 3rd,..., mth reading parameters in the nth historical information browsing record in any way;

[0067] For example, the historical information browsing record set of the first method is {the first browsing record, the second browsing record, the third browsing record};

[0068] Among them, the reading parameter set of the first browsing record is {the first reading parameter, the second reading parameter, the third reading parameter}.

[0069] Step S200: In the historical information browsing record set of any method, obtain the reading parameters after the user browses the information, calculate the reading index of the information, and determine whether the information is characteristic information;

[0070] Among them, step S200 includes:

[0071] Step S201: In the historical information browsing record set of any method, calculate the reading index of the nth historical information browsing record according to the formula:

[0072]

[0073] Among them, X bn represents the reading index of the nth historical information browsing record of the bth method, X bnc represents the value of the cth reading parameter in the nth historical information browsing record of the bth method, and d bc represents the weight of the cth reading parameter of the bth method;

[0074] Step S202: When X bn ≥X b where X b represents the reading index threshold of the bth method, it is determined that the nth historical information browsing record of the bth method is the characteristic information of the bth method, and the characteristic information of any method is summarized to obtain the characteristic information set of any method and the reading index set of the characteristic information.

[0075] Step S300: According to the information type of the information publishing platform system, confirm the type of the characteristic information, perform text processing on the characteristic information of any type, obtain the entry composition and entry proportion of the characteristic information, and calculate the weight of any entry;

[0076] Among them, step S300 includes:

[0077] Step S301: According to the information type of the information publishing platform, collect the information type of the characteristic information in the characteristic information set of any method, summarize a certain information type obtained by the collection, and obtain the characteristic information set of any information type;

[0078] Step S302: Input the feature information in the feature information set of any information type into the NLP model for text analysis, extract the entries of the feature information, summarize the number of times any entry appears in the feature information set, and calculate the frequency of a certain entry in any feature information according to the following formula:

[0079]

[0080] where Y tm represents the frequency of the m-th entry in the t-th feature information, and y tm represents the number of times the m-th entry appears in the t-th feature information, and y t represents the total number of times all entries appear in the t-th feature information;

[0081] Step S303: In the reading index set of the feature information, calculate the weight of a certain entry in any feature information according to the following formula:

[0082] Z tm = Y tm × X t ;

[0083] where Z tm represents the weight of the m-th entry in the t-th feature information, and X t represents the reading index of the t-th feature information, obtain the entry weight set of any feature information, and summarize the feature information of any information type to obtain the weight set of a certain entry in any information type;

[0084] For example, in the first method, the frequency of the first entry in the first feature information of the first information type is 0.3, the frequency of the second entry is 0.25, and the frequency of the third entry is 0.22;

[0085] The reading index of the first feature information is 3.8. Then, in the first method, the weight of the first entry in the first feature information of the first information type is 1.14, the weight of the second entry is 0.98, and the weight of the third entry is 0.836.

[0086] Step S400: Calculate the weight of any entry of any information type according to the weight of any entry in Step S300, and divide the level of the entry according to the weight;

[0087] where Step S400 includes:

[0088] Step S401: According to any method of obtaining information, divide the weight set of a certain entry of the same information type to obtain the weight set of a certain entry of any method in the same information type;

[0089] Step S402: Calculate the weight of a certain entry in any information type according to the following formula:

[0090]

[0091] where, Z pm represents the weight of the m-th entry in the p-th information type, and Z rpmi represents the i-th weight value of the m-th entry in the p-th information type in the r-th manner, and j rpm represents the number of weight value sets of the m-th entry in the p-th information type in the r-th manner, and α r represents the weight of the r-th manner, L represents the number of information acquisition manners, and obtain the entry weight set of any information type;

[0092] Step S403: Divide the entry weights into several levels, divide the entry weight set of any information type according to the levels, confirm the entry levels of any information type and summarize them to obtain an entry set of a certain level in any information type.

[0093] Step S500: Obtain the scoring record of the user for a certain historical browsing information. In the scoring record, form a data group with the score of the information by the user and the number of entries of any level, and perform function fitting on the data group to obtain a scoring prediction model for any information type;

[0094] where, Step S500 includes:

[0095] Step S501: Obtain the scoring record of the user for a certain historical browsing information, collect the information type and entries of the browsing information, determine the entry set of any level according to the collected information type, compare the collected entries with the entry set of any level, and obtain the number of entries of any level;

[0096] Step S502: Take the score of the information and the number of entries of any level as a data group, summarize the scoring records of the historical browsing information, and perform function fitting on several data groups according to the following formula:

[0097]

[0098] where, H represents the score of the information, S e represents the number of entries of the e-th level, β e represents the weight of the e-th level, f is the number of levels, S represents the total number of entries that meet any level in any information, S1 represents the total number of entries of any information, and γ represents the weight of the information score, and obtain the prediction score model of any information type.

[0099] Step S600: Sort the information to be released in real time according to the scoring prediction model, and push the information to users in the arranged order.

[0100] Among them, step S600 includes:

[0101] Step S601: Obtain the information type and entries of the information to be released in real time, determine the number of entries at any level that meet a certain user, and input the number of entries at the any level into the prediction scoring model of the same information type to obtain the user's prediction score.

[0102] Step S602: When the user's prediction score is greater than or equal to the scoring threshold, set the information to be released as interesting information, obtain a set of interesting information, sort the set according to the prediction scores of the interesting information from high to low, and push the sorted set of interesting information to the user.

[0103] To better implement the above method, an information release service system is also proposed. The system includes a historical information browsing record module, a feature information module, a term weight module, a term level module, a scoring prediction model module, and a real-time sorting module.

[0104] Historical information browsing record module: Through the information release platform system, obtain the user's historical information browsing records, collect the information acquisition methods of the historical information browsing records, divide the historical information browsing records according to different information acquisition methods, and obtain a set of historical information browsing records in any method.

[0105] Feature information module: In the set of historical information browsing records in any method, obtain the reading parameters of the user after browsing the information, calculate the reading index of the information through calculation, and determine whether the information is feature information.

[0106] Term weight module: According to the information type of the information release platform system, confirm the type of the feature information, perform text processing on the feature information of any type, obtain the term composition and term proportion of the feature information, and calculate the weight of any term.

[0107] Among them, the term weight module includes a feature information classification unit and a calculation term weight unit.

[0108] Feature information classification unit: According to the information type of the information release platform, collect the information types of the feature information, summarize a certain information type obtained by the collection, and obtain a set of feature information of any information type.

[0109] Calculation of entry weight unit: Through a natural language processing model, extract entries from the feature information, calculate the frequency of occurrence of an entry in the feature information, and calculate the weight of the entry based on the frequency and the reading index of the feature information.

[0110] Entry level module: Calculate the weight of any entry of any information type according to the weight of any entry in step S300, and divide the level of the entry according to the weight.

[0111] Among them, the entry level module includes a calculation of entry weight unit and a determination of level unit.

[0112] Calculation of entry weight unit: Obtain the set of weights of a certain entry in any way in the same information type, calculate the set of weights, and obtain the weight of a certain entry in any information type.

[0113] Determination of level unit: Divide the entry weights into several levels, divide the set of entry weights of any information type according to the levels, confirm and summarize the entry levels of any information type, and obtain the set of entries of a certain level in any information type.

[0114] Rating prediction model module: Obtain the rating record of a user for a certain historical browsing information. In the rating record, form a data group with the rating of the user for the information and the number of entries of any level, and perform function fitting on the data group to obtain the rating prediction model of any information type.

[0115] Real-time sorting module: Sort the information to be released in real time according to the rating prediction model, and push the information to the user in the arranged order.

[0116] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, for those skilled in the art, they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. An information publishing service method, characterized in that: Step S100: obtaining the user's historical information browsing records through the information publishing platform system, collecting the information acquisition methods of the historical information browsing records, dividing the historical information browsing records according to different information acquisition methods, and obtaining a set of historical information browsing records in any method; Step S200: in a set of historical information browsing records in any manner, obtaining reading parameters of the user after browsing the information, and obtaining the reading index of the information by calculation to determine whether the information is characteristic information; Step S300: confirm the type of the characteristic information according to the information type of the information publishing platform system, perform text processing on the characteristic information of any type, obtain the term composition and term proportion of the characteristic information, and calculate the weight of any term; Step S400: Calculate the weight of any term of any information type according to the weight of any term in step S300, and classify the terms according to the weight; The step S400 includes the following steps: Step S401: dividing the weight set of a certain term of the same information type according to any method of obtaining information, and obtaining the weight set of a certain term of any method in the same information type; Step S402: Calculate the weight of a term in any information type according to the following formula: ； Among them, Z pm It is expressed as the weight of the mth term in the pth information type, Z rpmi It is expressed as the i-th weight of the m-th term in the p-th information type in the r-th mode, j rpm It is expressed as the number of weight sets of the mth term in the pth information type in the rth way, α r It is represented as the weight of the rth way, L is represented as the number of ways to obtain information, and the weight set of terms of any information type is obtained; Step S403: Divide the term weights into several levels, divide the term weight set of any information type according to the level, confirm the term level of any information type and summarize it, and obtain a term set of a certain level in any information type; Step S500: obtaining a user's rating record for a certain historical browsing information, in which the user's rating of the information and the number of entries of any level are combined into a data group, and a function fitting is performed on the data group to obtain a rating prediction model for any information type; The step S500 includes the following steps: Step S501: obtaining a user's rating record for a certain historical browsing information, and collecting information types and entries of the browsing information, determining an entry set of any level according to the collected information type, comparing the collected entries with the entry set of any level, and obtaining the number of entries of any level; Step S502: The rating of the information and the number of entries of any level are regarded as a data group, the rating records of the historical browsing information are summarized, and a function fitting is performed on several data groups according to the following formula: ； Among them, H represents the score of information, S e Expressed as the number of terms at level e, β e It is represented as the weight of the e-th level, f is the number of levels, S is the total number of terms that meet any level in any information, S1 is the total number of terms in any information, and γ is the weight of the information score. The prediction score model of any information type is obtained; Step S600: sorting the real-time information to be published according to the rating prediction model, and pushing the information to the user in the sorted order.

2. The information publishing service method according to claim 1, characterized in that: In step S100, in the information publishing platform system, the user's historical information browsing records are obtained, and the historical information browsing records of the same method of obtaining information are summarized to obtain a historical information browsing record set A={a1, a2, a3, ..., an} of any method, wherein a1, a2, a3, ..., an respectively represent the 1st, 2nd, 3rd, ..., nth historical information browsing records of any method; wherein an={an1, an2, an3, ..., anm}, wherein an1, an2, an3, ..., anm respectively represent the values ​​of the 1st, 2nd, 3rd, ..., mth reading parameters in the nth historical information browsing record of any method.

3. The information publishing service method according to claim 2, characterized in that: The step S200 includes the following steps: Step S201: In any set of historical information browsing records, the reading index of the nth historical information browsing record is calculated according to the formula: ； Among them, X bn It is represented by the reading index of the nth historical information browsing record in the bth way, X bnc It is represented by the value of the reading parameter of item c in the browsing record of the nth historical information in the bth mode, d bc It is expressed as the weight of the cth reading parameter of the bth method; Step S202: When X bn ≥X b When X b The reading index threshold of the b-th mode is expressed as the reading index threshold of the b-th mode, and the nth historical information of the b-th mode is determined to be the characteristic information of the b-th mode. The characteristic information of any mode is aggregated to obtain a characteristic information set of any mode and a reading index set of the characteristic information.

4. The information publishing service method according to claim 3, characterized in that: The step S300 includes the following steps: Step S301: According to the information type of the information publishing platform, in the characteristic information set of any mode, the information type of the characteristic information is collected, and a certain information type collected is summarized to obtain the characteristic information set of any information type; Step S302: input the feature information in the feature information set of any information type into the NLP model for text analysis, extract the terms of the feature information, summarize the number of times any term appears in the feature information set, and calculate the frequency of a term in any feature information according to the following formula: ； Among them, Y tm It is represented by the frequency of the mth term in the tth feature information, y tm It is represented by the number of occurrences of the mth term in the tth feature information, y t It is represented by the number of occurrences of all terms in the t-th feature information; Step S303: In the reading index set of feature information, the weight of a certain term in any feature information is calculated according to the following formula: ； Among them, Z tm It is represented as the weight of the mth term in the tth feature information, X t It is expressed as the reading index of the t-th feature information, and the entry weight set of any feature information is obtained. The feature information of any information type is summarized to obtain the weight set of a certain entry in any information type.

5. The information publishing service method according to claim 4, characterized in that The step S600 includes the following steps: Step S601: obtaining the information type and terms of the real-time information to be published, determining the number of terms that meet the arbitrary level of a certain user, inputting the number of terms at the arbitrary level into a prediction and scoring model of the same information type, and obtaining the user's prediction score; Step S602: When the predicted score of the user is greater than or equal to the score threshold, the information to be released is set as information of interest, a set of information of interest is obtained, the set is sorted from high to low according to the predicted score of the information of interest, and the sorted set of information of interest is pushed to the user.

6. An information publishing service system, used to implement an information publishing service method as described in any one of claims 1 to 5, characterized in that: The system includes a historical information browsing record module, a feature information module, an entry weight module, an entry level module, a rating prediction model module, and a real-time sorting module; The historical information browsing record module: obtains the user's historical information browsing record through the information publishing platform system, collects the information acquisition method of the historical information browsing record, divides the historical information browsing record according to different information acquisition methods, and obtains a historical information browsing record set in any method; The characteristic information module: in any mode of historical information browsing record set, obtains the reading parameters of the user after browsing the information, and obtains the reading index of the information by calculation, and determines whether the information is characteristic information; The term weight module: confirms the type of the feature information according to the information type of the information publishing platform system, performs text processing on any type of feature information, obtains the term composition and term proportion of the feature information, and calculates the weight of any term; The term ranking module: calculates the weight of any term of any information type according to the weight of any term in step S300, and classifies the terms according to the weight; The rating prediction model module: obtains the user's rating record for a certain historical browsing information, in which the user's rating of the information and the number of entries of any level are combined into a data group, and a function is fitted on the data group to obtain a rating prediction model for any information type; The real-time sorting module sorts the real-time information to be published according to the rating prediction model, and pushes the information to the user in the sorted order.

7. An information publishing service system according to claim 6, characterized in that: The term weight module includes a feature information classification unit and a term weight calculation unit; The characteristic information classification unit: collects information types of characteristic information according to the information type of the information publishing platform, summarizes a certain information type collected, and obtains a characteristic information set of any information type; The term weight calculation unit extracts terms from feature information through a natural language processing model, calculates the frequency of terms appearing in the feature information, and calculates the weight of the terms based on the frequency and the reading index of the feature information.

8. An information publishing service system according to claim 6, characterized in that: The term ranking module includes a term weight calculation unit and a ranking determination unit; The term weight calculation unit: obtains a weight set of a term in any manner in the same information type, calculates the weight set, and obtains the weight of a term in any information type; The level determination unit divides the term weights into several levels, divides the term weight set of any information type according to the levels, confirms and summarizes the term level of any information type, and obtains a term set of a certain level in any information type.

Images