CN115757935A - Big data pushing method and system applying computer intelligence - Google Patents
Big data pushing method and system applying computer intelligence Download PDFInfo
- Publication number
- CN115757935A CN115757935A CN202211283411.7A CN202211283411A CN115757935A CN 115757935 A CN115757935 A CN 115757935A CN 202211283411 A CN202211283411 A CN 202211283411A CN 115757935 A CN115757935 A CN 115757935A
- Authority
- CN
- China
- Prior art keywords
- regression analysis
- secondary regression
- analysis unit
- basic
- behavior
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The method and the system for pushing the big data by applying the computer intelligence are realized based on a behavior interest mining model obtained by debugging a disturbance suppression rule when the pushing interest item is mined, the disturbance suppression rule can improve knowledge misleading caused by various unlimited disturbance element vectors possibly possessed under different interest subject knowledge, based on the debugging mode, the anti-interference performance of the debugged behavior interest mining model on random digital interconnection behavior disturbance data can be improved, better debugging quality and model operation expectation can be obtained without complex parameter adjustment, and the timeliness of model debugging can be improved. On the basis, the mining of the push interest items can be rapidly and accurately realized, so that the user interest or the user requirement is accurately adapted according to the push interest items to improve the efficiency of information push, and the waste of network resources caused by invalid push or low-efficiency push is reduced.
Description
Technical Field
The application relates to the technical field of big data pushing, in particular to a big data pushing method and system applying computer intelligence.
Background
Information push based on big data is a relatively complex technology, factors (dimensions) needing to be considered are very many, valuation of various information can be completed based on the big data technology at different stages (information collection, sorting and analysis), and on the basis, matching is performed with a user through an algorithm. The information-based push in the big data era has certain value. The 'best of the people' is taken as one of the characteristics and advantages of big data push, and the push precision and efficiency are always the key points concerned by all circles.
Disclosure of Invention
An object of the present application is to provide a big data pushing method and system applying computer intelligence.
A big data pushing method applying computer intelligence is applied to a big data pushing system, and comprises the following steps:
mining a push interest item in the digital interconnection behavior big data to be analyzed through a behavior interest mining model to obtain mining information of the push interest item, wherein the mining information comprises an interest subject of the push interest item, and the behavior interest mining model is obtained by debugging a disturbance suppression rule;
and performing information push of the digital interconnection client aiming at the analog analysis digital interconnection behavior big data based on the mining information.
Optionally, the step of debugging the behavioral interest mining model through a perturbation suppression rule includes:
performing behavior element mining on a plurality of authenticated digital interconnection behavior big data carrying the same interest topic knowledge in an authenticated model tuning set through a behavior interest mining model to obtain a behavior element vector relation network under the same interest topic knowledge, wherein the authenticated model tuning set comprises the authenticated digital interconnection behavior big data carrying the plurality of interest topic knowledge;
setting a plurality of basic secondary regression analysis units for the behavior element vector relationship network, and determining the basic secondary regression analysis units matched with the behavior element vectors in the behavior element vector relationship network respectively;
optimizing the plurality of basic secondary regression analysis units by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same interest topic knowledge;
determining model quality cost according to the target secondary regression analysis units under the knowledge of various interest subjects and the behavior element vectors matched with the target secondary regression analysis units, and debugging the behavior interest mining model according to the model quality cost.
Optionally, the optimizing the multiple basic secondary regression analysis units by combining the common scores between the multiple behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same subject of interest knowledge includes:
judging whether island element vectors exist in the behavior element vectors or not by combining the common scores between the behavior element vectors and the respectively matched basic secondary regression analysis units, wherein the common score between the island element vectors and the corresponding basic secondary regression analysis units is smaller than a first limit value;
on the basis that island element vectors exist in the plurality of behavior element vectors, establishing a basic secondary regression analysis unit by combining the island element vectors, and optimizing the island element vectors to belong to the established basic secondary regression analysis unit; the target secondary regression analysis unit comprises a created basic secondary regression analysis unit and a plurality of set basic secondary regression analysis units.
Optionally, the determining, by combining the common scores between the multiple behavior element vectors and the respectively matched basic secondary regression analysis units, whether an island element vector exists in the multiple behavior element vectors includes:
for any basic secondary regression analysis unit, determining a first limit value corresponding to the basic secondary regression analysis unit by combining a first evaluation index and a second evaluation index of a common score between the basic secondary regression analysis unit and a behavior element vector belonging to the basic secondary regression analysis unit;
judging whether behavior element vectors with the commonality score smaller than the first limit value exist in the behavior element vectors belonging to the basic secondary regression analysis unit or not;
and determining the existing behavior element vector with the common score smaller than the first limit value with the basic secondary regression analysis unit as the island element vector not matched with the basic secondary regression analysis unit.
Optionally, the optimizing the multiple basic secondary regression analysis units by combining the common scores between the multiple behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same subject of interest knowledge includes:
judging whether an abnormal secondary regression analysis unit exists in the plurality of basic secondary regression analysis units or not by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units, wherein a first evaluation index of the common score between the behavior element vector subordinate to the abnormal secondary regression analysis unit and the abnormal secondary regression analysis unit is not more than a second limit value;
on the basis that an abnormal secondary regression analysis unit exists in the plurality of basic secondary regression analysis units, cleaning the abnormal secondary regression analysis unit and the behavior element vector belonging to the abnormal secondary regression analysis unit to obtain a reserved basic secondary regression analysis unit and the behavior element vector belonging to the reserved basic secondary regression analysis unit; wherein the target secondary regression analysis unit comprises the retained base secondary regression analysis unit.
Optionally, the determining, by combining the common scores between the behavior element vectors and the respectively matched basic secondary regression analysis units, whether there is an abnormal secondary regression analysis unit in the multiple basic secondary regression analysis units includes:
for any basic secondary regression analysis unit, judging whether a first evaluation index of the commonality score between the basic secondary regression analysis unit and the behavior element vector belonging to the basic secondary regression analysis unit is not more than a second limit value;
and on the basis that the first evaluation index of the commonality score between the basic secondary regression analysis unit and the behavior element vector belonging to the basic secondary regression analysis unit is not more than the second limit value, determining that the basic secondary regression analysis unit is an abnormal secondary regression analysis unit.
Optionally, the optimizing the multiple basic secondary regression analysis units by combining the common scores between the multiple behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same subject of interest knowledge includes: determining at least one group of to-be-combined secondary regression analysis units in the plurality of basic secondary regression analysis units by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units, wherein each group of to-be-combined secondary regression analysis units comprises at least two basic secondary regression analysis units meeting a common relation; clustering the at least one group of secondary regression analysis units to be combined respectively to obtain at least one clustered secondary regression analysis unit, and optimizing the behavior element vectors belonging to the groups of secondary regression analysis units to be combined into clustered secondary regression analysis units; wherein the target secondary regression analysis unit comprises the clustered secondary regression analysis unit;
wherein the determining at least one group of to-be-combined secondary regression analysis units in the plurality of basic secondary regression analysis units by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units comprises: determining a third limit value corresponding to each basic secondary regression analysis unit according to the first evaluation index and the second evaluation index of the common score between each basic secondary regression analysis unit and the behavior element vector matched with each basic secondary regression analysis unit; for any basic secondary regression analysis unit, judging whether the common score between the basic secondary regression analysis unit and the rest basic secondary regression analysis units in the plurality of basic secondary regression analysis units is not less than a fourth limit value, wherein the fourth limit value is the maximum value of a third limit value corresponding to the basic secondary regression analysis unit and a third limit value corresponding to the rest basic secondary regression analysis units; determining that the residual basic secondary regression analysis unit and the basic secondary regression analysis unit satisfy a common relationship on the basis that the common score between the basic secondary regression analysis unit and the residual basic secondary regression analysis unit is not less than the maximum value; and determining the basic secondary regression analysis unit and at least one residual basic secondary regression analysis unit which meets the common relation with the basic secondary regression analysis unit as a group of secondary regression analysis units to be combined.
Optionally, the determining the model quality cost according to the target secondary regression analysis units under the knowledge of the multiple interest topics and the behavior element vectors matched with the target secondary regression analysis units includes:
for each behavior element vector subordinate to any target secondary regression analysis unit, determining a local model quality cost corresponding to the behavior element vector according to a common score between the behavior element vector and the subordinate target secondary regression analysis unit and common scores between the behavior element vector and remaining target secondary regression analysis units, wherein the remaining target secondary regression analysis units comprise target secondary regression analysis units which are not included in the target secondary regression analysis units under the knowledge of the multiple interest topics, except the target secondary regression analysis unit to which the behavior element vector belongs, and have a common score with the behavior element vector not greater than a fifth limit value;
and determining the model quality cost according to the local model quality cost corresponding to each behavior element vector matched with each target secondary regression analysis unit.
Optionally, the determining a basic secondary regression analysis unit that the plurality of behavior element vectors in the behavior element vector relationship network are respectively matched includes: and determining the common score between the behavior element vector and each basic secondary regression analysis unit aiming at any behavior element vector in the behavior element vector relation network, and determining the basic secondary regression analysis unit corresponding to the maximum common score as the basic secondary regression analysis unit to which the behavior element vector belongs.
A big data push system, comprising: a memory for storing an executable computer program, a processor for implementing the above method when executing the executable computer program stored in the memory.
A computer-readable storage medium, on which a computer program is stored which, when executed, performs the above-described method.
According to the method and the system for pushing the big data by applying the computer intelligence, when the pushing interest item is mined, the pushing interest item is realized based on the behavior interest mining model obtained by debugging the disturbance suppression rule, the disturbance suppression rule can improve knowledge misleading caused by various unlimited disturbance element vectors which may be possessed under different interest subject knowledge, based on the debugging mode, the anti-interference performance of the debugged behavior interest mining model on random digital interconnection behavior disturbance data can be improved, better debugging quality and model operation expectation can be obtained without complex parameter adjustment, and the timeliness of model debugging can be improved. On the basis, the mining of the push interest items can be rapidly and accurately realized, so that the user interest or the user requirement is accurately adapted according to the push interest items to improve the efficiency of information push, and the waste of network resources caused by invalid push or low-efficiency push is reduced.
In addition, a plurality of basic secondary regression analysis units are arranged for the behavior element vector relationship network under the same interest topic knowledge, the basic secondary regression analysis units matched with the behavior element vectors in the behavior element vector relationship network are determined, and a plurality of secondary regression analysis units are equivalently arranged for the behavior element vectors under the same interest topic knowledge so as to collect disturbance element vectors possibly possessed under the same interest topic knowledge; optimizing the multiple basic secondary regression analysis units according to the commonality scores between the multiple behavior element vectors and the respectively matched basic secondary regression analysis units, so that the obtained at least one target secondary regression analysis unit under the same interest topic knowledge can be intelligently matched with the real incidence relation of the interest topic knowledge downlink as element vectors as much as possible, namely the behavior element vectors under the same interest topic knowledge can be clustered to the target secondary regression analysis units as accurate as possible, and the misleading of knowledge generated by various unlimited disturbance element vectors possibly possessed under different interest topic knowledge is improved; and then, the behavior interest mining model is debugged through the model quality cost determined by the target secondary regression analysis unit and the behavior element vector belonging to the target secondary regression analysis unit, so that the anti-interference performance of the debugged behavior interest mining model on random digital interconnection behavior disturbance data can be improved in the debugging process of debugging the behavior interest mining model through an authenticated model tuning set which possibly contains the digital interconnection behavior disturbance data, better debugging quality and model operation expectation can be obtained without complex parameter adjustment, and the timeliness of model debugging can be improved.
Drawings
Fig. 1 is a schematic diagram showing one communication configuration of a big data push system in which an embodiment of the present application can be implemented.
Fig. 2 is a flow chart diagram illustrating a big data pushing method applying computer intelligence, which can implement an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be described in further detail with reference to the attached drawings, the described embodiments should not be considered as limiting the present application, and all other embodiments obtained by a person of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application. In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of the present application only and is not intended to be limiting of the application.
Fig. 1 is a block diagram illustrating a communication configuration of a big data pushing system 100 that can implement an embodiment of the present application, where the big data pushing system 100 includes a memory 101 for storing an executable computer program, and a processor 102 for implementing a big data pushing method applying computer intelligence in the embodiment of the present application when executing the executable computer program stored in the memory 101.
Fig. 2 is a flowchart illustrating a big data pushing method using computer intelligence, which may be implemented by the big data pushing system 100 shown in fig. 1, and further may include the technical solutions described in S1 and S2 below.
S1, mining the push interest items in the digital interconnection behavior big data to be analyzed through a behavior interest mining model to obtain mining information of the push interest items.
In this embodiment of the present application, the mining information includes the interest topic of the pushed interest item, where the behavioral interest mining model is obtained by debugging a perturbation suppression rule, and the behavioral interest mining model may be a convolutional neural network, a deep learning model, or a multi-class model, where the type and architecture of the behavioral interest mining model are not limited as long as the technical solution can be implemented. For example, the digital interconnection behavior big data to be analyzed can be digital interconnection behavior big data to be processed, and relates to meta universe service, digital cloud service, e-commerce service, intelligent government and enterprise service and the like. The push interest item can be a push interest point or a push interest event of a user in the service interaction process, and the interest topic can be understood as an interest category for distinguishing the push interest item.
And S2, performing information pushing of the digital interconnection client aiming at the analog analysis digital interconnection behavior big data based on the mining information.
It can be understood that after the mining information is obtained, the data information to be pushed can be matched according to the mining information, so that the user interest or the user requirement can be accurately adapted when the information is pushed to the digital interconnection client, the information pushing efficiency is improved, and the network resource waste caused by invalid pushing or low-efficiency pushing is reduced.
The method is applied to S1 and S2, and is realized based on the behavior interest mining model obtained by debugging the disturbance suppression rule when the pushing interest item is mined, the disturbance suppression rule can improve knowledge misleading caused by various unlimited disturbance element vectors possibly under different interest subject knowledge, the anti-interference performance of the debugged behavior interest mining model on random digital interconnection behavior disturbance data can be improved based on the debugging mode, better debugging quality and model operation expectation can be obtained without complex parameter adjustment, and the timeliness of model debugging can be improved. On the basis, the mining of the push interest items can be rapidly and accurately realized, so that the user interest or the user requirement is accurately adapted according to the push interest items to improve the efficiency of information push, and the waste of network resources caused by invalid push or low-efficiency push is reduced.
In other possible embodiments, the step of debugging the behavioral interest mining model by using the disturbance suppression rule includes performing behavioral element mining on a plurality of authenticated digital interconnected behavior big data (digital interconnected behavior big data samples) carrying the same interest topic knowledge (interest category labels) in an authenticated model tuning set (training sample set) by using the behavioral interest mining model to obtain a behavioral element vector relationship network (sample feature set) under the same interest topic knowledge, wherein the authenticated model tuning set comprises the authenticated digital interconnected behavior big data carrying a plurality of interest topic knowledge; setting a plurality of basic secondary regression analysis units (initial clustering centers, which can be understood as subclass centers) for the behavior element vector relationship network, and determining the basic secondary regression analysis units respectively matched with a plurality of behavior element vectors in the behavior element vector relationship network; optimizing (adjusting) the plurality of basic secondary regression analysis units by combining the common scores (feature similarity) between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same interest topic knowledge; determining model quality cost (model loss function) according to the target secondary regression analysis units under the knowledge of various interest subjects and the behavior element vectors (user behavior characteristic samples) matched with the target secondary regression analysis units, and debugging the behavior interest mining model according to the model quality cost.
It can be understood that, by setting a plurality of basic secondary regression analysis units for the behavior element vector relationship network under the same interest topic knowledge, and determining the basic secondary regression analysis units respectively matched with a plurality of behavior element vectors in the behavior element vector relationship network, it is equivalent to setting a plurality of secondary regression analysis units for a plurality of behavior element vectors under the same interest topic knowledge, so as to collect disturbance element vectors that may be possessed under the same interest topic knowledge; optimizing the multiple basic secondary regression analysis units according to the commonality scores between the multiple behavior element vectors and the respectively matched basic secondary regression analysis units, so that the obtained at least one target secondary regression analysis unit under the same interest topic knowledge can be intelligently matched with the real incidence relation of the interest topic knowledge downlink as the element vector as much as possible, namely the behavior element vectors under the same interest topic knowledge can be clustered to the target secondary regression analysis unit which is as accurate as possible, thereby improving the knowledge misleading (label misleading) generated by various unlimited disturbance element vectors which may be possessed under different interest topic knowledge and realizing the noise suppression processing; and then, the behavior interest mining model is debugged through the model quality cost determined by the target secondary regression analysis unit and the behavior element vector belonging to the target secondary regression analysis unit, so that the anti-interference performance of the debugged behavior interest mining model on random digital interconnection behavior disturbance data can be improved in the debugging process of debugging the behavior interest mining model through an authenticated model tuning set which possibly contains the digital interconnection behavior disturbance data, better debugging quality and model operation expectation can be obtained without complex parameter adjustment, and the timeliness of model debugging can be improved.
Under other possible design ideas, the optimizing the multiple basic secondary regression analysis units by combining the commonality scores between the multiple behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same interest topic knowledge includes: judging whether island element vectors (relatively isolated feature vectors or cluster members without aggregation relation) exist in the behavior element vectors by combining the commonality scores between the behavior element vectors and the respectively matched basic secondary regression analysis units, wherein the commonality score between the island element vector and the corresponding basic secondary regression analysis unit is smaller than a first limit value; on the basis that island element vectors exist in the plurality of behavior element vectors, establishing a basic secondary regression analysis unit by combining the island element vectors, and optimizing the island element vectors to belong to the established basic secondary regression analysis unit; the target secondary regression analysis unit comprises a created basic secondary regression analysis unit and a plurality of set basic secondary regression analysis units.
In this way, the integrity of the obtained at least one target secondary regression analysis unit under the knowledge of the same topic of interest can be ensured.
For some possible design ideas, the determining whether an island element vector exists in the behavior element vectors by combining the common scores between the behavior element vectors and the respectively matched basic secondary regression analysis units includes: for any basic secondary regression analysis unit, determining a first limit value corresponding to the basic secondary regression analysis unit by combining a first evaluation index (common score mean) and a second evaluation index (common score standard deviation) of common score between the basic secondary regression analysis unit and the behavior element vector belonging to the basic secondary regression analysis unit; judging whether behavior element vectors with the commonality score smaller than the first limit value exist in the behavior element vectors belonging to the basic secondary regression analysis unit or not; and determining the existing behavior element vector with the common score smaller than the first limit value with the basic secondary regression analysis unit as the island element vector which does not match with the basic secondary regression analysis unit. The limit value can be understood as a threshold value or a preset value.
In this way, by introducing the mean index and the standard deviation index for analysis, whether the plurality of behavior element vectors have the island element vector can be accurately judged.
In some exemplary embodiments, the optimizing the plurality of basic secondary regression analysis units by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same subject of interest knowledge includes: judging whether an abnormal secondary regression analysis unit exists in the plurality of basic secondary regression analysis units or not by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units, wherein a first evaluation index of the common score between the behavior element vector subordinate to the abnormal secondary regression analysis unit and the abnormal secondary regression analysis unit is not more than a second limit value; on the basis that an abnormal secondary regression analysis unit exists in the plurality of basic secondary regression analysis units, cleaning the abnormal secondary regression analysis unit and the behavior element vector belonging to the abnormal secondary regression analysis unit to obtain a reserved basic secondary regression analysis unit and the behavior element vector belonging to the reserved basic secondary regression analysis unit; wherein the target secondary regression analysis unit comprises the retained base secondary regression analysis unit.
By the design, the abnormal secondary regression analysis unit and the behavior element vector belonging to the abnormal secondary regression analysis unit can be cleaned, so that the precision and the signal-to-noise ratio of at least one target secondary regression analysis unit under the same interest subject knowledge are improved.
In other possible embodiments, the determining whether there is an abnormal secondary regression analysis unit in the multiple basic secondary regression analysis units by combining the commonality scores between the multiple behavior element vectors and the respectively matched basic secondary regression analysis units includes: for any basic secondary regression analysis unit, judging whether a first evaluation index of the common score between the basic secondary regression analysis unit and the behavior element vector belonging to the basic secondary regression analysis unit is not larger than a second limit value or not; and on the basis that the first evaluation index of the commonality score between the basic secondary regression analysis unit and the behavior element vector belonging to the basic secondary regression analysis unit is not more than the second limit value, determining that the basic secondary regression analysis unit is an abnormal secondary regression analysis unit.
In this way, the abnormal secondary regression analysis unit can be accurately identified based on the mean index.
In some examples, the optimizing the plurality of basic secondary regression analysis units in combination with the commonality scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same subject of interest knowledge includes: determining at least one group of to-be-combined secondary regression analysis units in the plurality of basic secondary regression analysis units by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units, wherein each group of to-be-combined secondary regression analysis units comprises at least two basic secondary regression analysis units meeting a common relation (similarity or similarity); clustering (merging or combining) is carried out on the at least one group of secondary regression analysis units to be combined respectively to obtain at least one clustered secondary regression analysis unit, and the behavior element vectors belonging to the groups of secondary regression analysis units to be combined are optimized to belong to the clustered secondary regression analysis units; wherein the target secondary regression analysis unit comprises the clustered secondary regression analysis unit.
By the design, the optimization adjustment of the target secondary regression analysis unit can be realized based on clustering processing, so that the signal-to-noise ratio of the target secondary regression analysis unit is improved, and noise interference is reduced.
In some possible embodiments, the determining at least one group of secondary regression analysis units to be combined in the plurality of basic secondary regression analysis units in combination with the common score between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units includes: determining a third limit value corresponding to each basic secondary regression analysis unit according to the first evaluation index and the second evaluation index of the common score between each basic secondary regression analysis unit and the behavior element vector matched with each basic secondary regression analysis unit; for any basic secondary regression analysis unit, judging whether the commonality score between the basic secondary regression analysis unit and the rest basic secondary regression analysis units in the plurality of basic secondary regression analysis units is not less than a fourth limit value, wherein the fourth limit value is the maximum value of the third limit value corresponding to the basic secondary regression analysis unit and the third limit value corresponding to the rest basic secondary regression analysis units; determining that the residual basic secondary regression analysis unit and the basic secondary regression analysis unit satisfy a common relationship on the basis that the common score between the basic secondary regression analysis unit and the residual basic secondary regression analysis unit is not less than the maximum value; and determining the basic secondary regression analysis unit and at least one residual basic secondary regression analysis unit which meets the common relation with the basic secondary regression analysis unit as a group of secondary regression analysis units to be combined.
In this way, the secondary regression analysis unit to be combined can be accurately and reliably determined based on the analysis of the commonality scores under different limit judgment conditions.
Under some design ideas which can be independently implemented, determining the model quality cost according to the target secondary regression analysis units under the knowledge of various interest topics and the behavior element vectors matched with the target secondary regression analysis units comprises: for each behavior element vector subordinate to any target secondary regression analysis unit, determining a local model quality cost (model sub-loss) corresponding to the behavior element vector according to a commonality score between the behavior element vector and the subordinate target secondary regression analysis unit and commonality scores between the behavior element vector and remaining target secondary regression analysis units, wherein the remaining target secondary regression analysis units comprise target secondary regression analysis units, except the target secondary regression analysis unit to which the behavior element vector belongs, in the plurality of target secondary regression analysis units under the knowledge of the plurality of interest topics, and the commonality score between the target secondary regression analysis units and the behavior element vector is not greater than a fifth limit value; and determining the model quality cost according to the local model quality cost corresponding to each behavior element vector matched with each target secondary regression analysis unit.
It can be understood that the model quality cost can be determined comprehensively and accurately by considering different local model quality costs.
Under some design ideas which can be independently implemented, the basic secondary regression analysis unit for determining that a plurality of behavior element vectors in the behavior element vector relationship network are respectively matched comprises: and determining the common score between the behavior element vector and each basic secondary regression analysis unit aiming at any behavior element vector in the behavior element vector relation network, and determining the basic secondary regression analysis unit corresponding to the maximum common score as the basic secondary regression analysis unit to which the behavior element vector belongs.
The embodiments of the present application have been described above with reference to the accompanying drawings, and have at least the following beneficial effects: when the push interest item is mined, the push interest item mining is realized based on the behavior interest mining model obtained by debugging the disturbance suppression rule, the disturbance suppression rule can improve knowledge misleading caused by various unlimited disturbance element vectors which may be possessed under different interest subject knowledge, based on the debugging mode, the anti-interference performance of the debugged behavior interest mining model on random digital interconnection behavior disturbance data can be improved, better debugging quality and model operation expectation can be obtained without complex parameter adjustment, and the timeliness of model debugging can be improved. On the basis, the mining of the push interest items can be rapidly and accurately realized, so that the user interest or the user requirement is accurately adapted according to the push interest items to improve the efficiency of information push, and the waste of network resources caused by invalid push or low-efficiency push is reduced.
In addition, a plurality of basic secondary regression analysis units are arranged for the behavior element vector relationship network under the same interest topic knowledge, the basic secondary regression analysis units matched with a plurality of behavior element vectors in the behavior element vector relationship network are determined, and a plurality of secondary regression analysis units are equivalently arranged for the plurality of behavior element vectors under the same interest topic knowledge so as to collect disturbance element vectors possibly possessed under the same interest topic knowledge; optimizing the basic secondary regression analysis units according to the commonality scores between the behavior element vectors and the basic secondary regression analysis units which are respectively matched with the behavior element vectors, so that the obtained at least one target secondary regression analysis unit under the same interest topic knowledge can be intelligently matched with the real incidence relation of the interest topic knowledge downlink as the element vector as much as possible, namely the behavior element vectors under the same interest topic knowledge can be clustered to the target secondary regression analysis unit which is as accurate as possible, thereby improving the knowledge misleading caused by various unrestrained disturbance element vectors possibly possessed under different interest topic knowledge; and debugging the behavior interest mining model through model quality cost determined by the target secondary regression analysis unit and the behavior element vector belonging to the target secondary regression analysis unit, so that the anti-interference performance of the debugged behavior interest mining model on random digital interconnection behavior disturbance data can be improved in the debugging process of debugging the behavior interest mining model through an authenticated model tuning set which possibly comprises the digital interconnection behavior disturbance data, better debugging quality and model operation expectation can be obtained without complex parameter adjustment, and the timeliness of model debugging can be improved.
The above description is only a preferred embodiment of the present application, and is not intended to limit the scope of the present application.
Claims (10)
1. A big data pushing method applying computer intelligence is characterized in that the method is applied to a big data pushing system, and the method at least comprises the following steps:
mining a push interest item in the digital interconnection behavior big data to be analyzed through a behavior interest mining model to obtain mining information of the push interest item, wherein the mining information comprises an interest subject of the push interest item, and the behavior interest mining model is obtained by debugging a disturbance suppression rule;
and performing information push of the digital interconnection client aiming at the analog analysis digital interconnection behavior big data based on the mining information.
2. The method of claim 1, wherein the step of tuning the behavioral interest mining model via perturbation suppression rules comprises:
performing behavior element mining on a plurality of authenticated digital interconnection behavior big data carrying the same interest topic knowledge in an authenticated model tuning set through a behavior interest mining model to obtain a behavior element vector relation network under the same interest topic knowledge, wherein the authenticated model tuning set comprises the authenticated digital interconnection behavior big data carrying the plurality of interest topic knowledge;
setting a plurality of basic secondary regression analysis units for the behavior element vector relationship network, and determining the basic secondary regression analysis units respectively matched with a plurality of behavior element vectors in the behavior element vector relationship network;
optimizing the plurality of basic secondary regression analysis units by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same interest topic knowledge;
determining model quality cost according to target secondary regression analysis units under multiple interest topic knowledge and behavior element vectors matched with each target secondary regression analysis unit, and debugging the behavior interest mining model according to the model quality cost.
3. The method according to claim 2, wherein the optimizing the plurality of basic secondary regression analysis units in combination with the common score between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same subject of interest knowledge comprises:
judging whether island element vectors exist in the behavior element vectors or not by combining the common scores between the behavior element vectors and the respectively matched basic secondary regression analysis units, wherein the common score between the island element vectors and the corresponding basic secondary regression analysis units is smaller than a first limit value;
on the basis that island element vectors exist in the plurality of behavior element vectors, establishing a basic secondary regression analysis unit by combining the island element vectors, and optimizing the island element vectors to belong to the established basic secondary regression analysis unit; the target secondary regression analysis unit comprises a created basic secondary regression analysis unit and a plurality of set basic secondary regression analysis units.
4. The method of claim 3, wherein the determining whether an islanding element vector exists in the plurality of behavior element vectors in combination with the commonality scores between the plurality of behavior element vectors and the respectively matched underlying secondary regression analysis units comprises:
for any basic secondary regression analysis unit, determining a first limit value corresponding to the basic secondary regression analysis unit by combining a first evaluation index and a second evaluation index of a common score between the basic secondary regression analysis unit and a behavior element vector belonging to the basic secondary regression analysis unit;
judging whether behavior element vectors with the common score smaller than the first limit value with the basic secondary regression analysis unit exist in the behavior element vectors belonging to the basic secondary regression analysis unit or not;
and determining the existing behavior element vector with the common score smaller than the first limit value with the basic secondary regression analysis unit as the island element vector which does not match with the basic secondary regression analysis unit.
5. The method according to any one of claims 2 to 4, wherein the optimizing the plurality of basic secondary regression analysis units in combination with the commonality scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same subject of interest knowledge comprises:
judging whether an abnormal secondary regression analysis unit exists in the plurality of basic secondary regression analysis units or not by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units, wherein a first evaluation index of the common score between the behavior element vector subordinate to the abnormal secondary regression analysis unit and the abnormal secondary regression analysis unit is not more than a second limit value;
on the basis that an abnormal secondary regression analysis unit exists in the plurality of basic secondary regression analysis units, cleaning the abnormal secondary regression analysis unit and the behavior element vector belonging to the abnormal secondary regression analysis unit to obtain a reserved basic secondary regression analysis unit and the behavior element vector belonging to the reserved basic secondary regression analysis unit; wherein the target secondary regression analysis unit comprises the retained base secondary regression analysis unit.
6. The method according to claim 5, wherein the determining whether an abnormal secondary regression analysis unit exists in the plurality of basic secondary regression analysis units by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units comprises:
for any basic secondary regression analysis unit, judging whether a first evaluation index of the common score between the basic secondary regression analysis unit and the behavior element vector belonging to the basic secondary regression analysis unit is not larger than a second limit value or not;
and on the basis that the first evaluation index of the commonality score between the basic secondary regression analysis unit and the behavior element vector belonging to the basic secondary regression analysis unit is not greater than the second limit value, determining the basic secondary regression analysis unit as an abnormal secondary regression analysis unit.
7. The method according to claim 6, wherein the optimizing the plurality of basic secondary regression analysis units in combination with the commonality scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units to obtain at least one target secondary regression analysis unit under the same subject of interest knowledge comprises: determining at least one group of to-be-combined secondary regression analysis units in the plurality of basic secondary regression analysis units by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units, wherein each group of to-be-combined secondary regression analysis units comprises at least two basic secondary regression analysis units meeting a common relation; clustering the at least one group of secondary regression analysis units to be combined respectively to obtain at least one clustered secondary regression analysis unit, and optimizing the behavior element vectors belonging to the groups of secondary regression analysis units to be combined into clustered secondary regression analysis units; wherein the target secondary regression analysis unit comprises the clustered secondary regression analysis unit;
wherein the determining at least one group of to-be-combined secondary regression analysis units in the plurality of basic secondary regression analysis units by combining the common scores between the plurality of behavior element vectors and the respectively matched basic secondary regression analysis units comprises: determining a third limit value corresponding to each basic secondary regression analysis unit according to the first evaluation index and the second evaluation index of the common score between each basic secondary regression analysis unit and the behavior element vector matched with each basic secondary regression analysis unit; for any basic secondary regression analysis unit, judging whether the common score between the basic secondary regression analysis unit and the rest basic secondary regression analysis units in the plurality of basic secondary regression analysis units is not less than a fourth limit value, wherein the fourth limit value is the maximum value of a third limit value corresponding to the basic secondary regression analysis unit and a third limit value corresponding to the rest basic secondary regression analysis units; determining that the residual basic secondary regression analysis unit and the basic secondary regression analysis unit satisfy a common relationship on the basis that the common score between the basic secondary regression analysis unit and the residual basic secondary regression analysis unit is not less than the maximum value; and determining the basic secondary regression analysis unit and at least one residual basic secondary regression analysis unit which meets the common relation with the basic secondary regression analysis unit as a group of secondary regression analysis units to be combined.
8. The method of claim 2, wherein determining the model quality cost according to the target secondary regression analysis units under the knowledge of the interest topics and the behavior element vectors matched with the target secondary regression analysis units comprises:
for each behavior element vector subordinate to any target secondary regression analysis unit, determining a local model quality cost corresponding to the behavior element vector according to a common score between the behavior element vector and the subordinate target secondary regression analysis unit and common scores between the behavior element vector and remaining target secondary regression analysis units, wherein the remaining target secondary regression analysis units comprise target secondary regression analysis units which are not included in the target secondary regression analysis units under the knowledge of the multiple interest topics, except the target secondary regression analysis unit to which the behavior element vector belongs, and have a common score with the behavior element vector not greater than a fifth limit value;
and determining the model quality cost according to the local model quality cost corresponding to each behavior element vector matched with each target secondary regression analysis unit.
9. The method of claim 2, wherein determining the base secondary regression analysis unit that the behavior element vectors in the behavior element vector relationship network respectively match comprises: and determining the common score between the behavior element vector and each basic secondary regression analysis unit aiming at any behavior element vector in the behavior element vector relation network, and determining the basic secondary regression analysis unit corresponding to the maximum common score as the basic secondary regression analysis unit to which the behavior element vector belongs.
10. A big data push system, comprising: a memory for storing an executable computer program, a processor for implementing the method of any one of claims 1-9 when executing the executable computer program stored in the memory.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211283411.7A CN115757935A (en) | 2022-10-20 | 2022-10-20 | Big data pushing method and system applying computer intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211283411.7A CN115757935A (en) | 2022-10-20 | 2022-10-20 | Big data pushing method and system applying computer intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115757935A true CN115757935A (en) | 2023-03-07 |
Family
ID=85353959
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211283411.7A Pending CN115757935A (en) | 2022-10-20 | 2022-10-20 | Big data pushing method and system applying computer intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115757935A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116578781A (en) * | 2023-04-28 | 2023-08-11 | 北京天译科技有限公司 | Meteorological service pushing method and server applying neural network algorithm |
-
2022
- 2022-10-20 CN CN202211283411.7A patent/CN115757935A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116578781A (en) * | 2023-04-28 | 2023-08-11 | 北京天译科技有限公司 | Meteorological service pushing method and server applying neural network algorithm |
| CN116578781B (en) * | 2023-04-28 | 2023-10-24 | 北京天译科技有限公司 | Meteorological service pushing method and server applying neural network algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112597141B (en) | Network flow detection method based on public opinion analysis | |
| CN111159404B (en) | Text classification method and device | |
| CN109657063A (en) | A kind of processing method and storage medium of magnanimity environment-protection artificial reported event data | |
| CN107704868A (en) | Tenant group clustering method based on Mobile solution usage behavior | |
| WO2020024444A1 (en) | Group performance grade recognition method and apparatus, and storage medium and computer device | |
| CN113449011A (en) | Big data prediction-based information push updating method and big data prediction system | |
| CN112163096A (en) | Malicious group determination method and device, electronic equipment and storage medium | |
| CN114048318A (en) | Clustering method, system, device and storage medium based on density radius | |
| CN115757935A (en) | Big data pushing method and system applying computer intelligence | |
| CN113486983A (en) | Big data office information analysis method and system for anti-fraud processing | |
| CN115294397A (en) | Classification task post-processing method, device, equipment and storage medium | |
| CN108959577B (en) | Entity matching method and computer program based on non-dominant attribute outlier detection | |
| CN110689359A (en) | Method and device for dynamically updating model | |
| CN116887201B (en) | Intelligent short message pushing method and system based on user analysis | |
| CN112925964A (en) | Big data acquisition method based on cloud computing service and big data acquisition service system | |
| CN108829846A (en) | A kind of business recommended platform data cluster optimization system and method based on user characteristics | |
| KR20200143989A (en) | The person in charge assignment system and method based on type classification for customer inquiries | |
| CN114610758A (en) | Data processing method and device based on data warehouse, readable medium and equipment | |
| CN112486676B (en) | Data sharing and distributing device based on edge calculation | |
| CN112463964B (en) | Text classification and model training method, device, equipment and storage medium | |
| CN114661998A (en) | Big data processing method and system based on Internet hot topics | |
| CN109308565B (en) | Crowd performance grade identification method and device, storage medium and computer equipment | |
| CN114780695A (en) | Big data mining method and big data mining system for online topics | |
| CN112925831A (en) | Big data mining method and big data mining service system based on cloud computing service | |
| CN116993307B (en) | Collaborative office method and system with artificial intelligence learning capability |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |