CN104978416A - Redis-based intelligent object retrieval method - Google Patents

Abstract

The invention discloses a Redis based intelligent object retrieval method. An object is analyzed; all retrieval persons retrieving the object within a time period are viewed; and time sequence sorting is carried out according to assigned weight or a set arrangement method by utilizing Sort-Set and List data structures of a Redis system through analyzing the retrieval persons to obtain recommended objects for the retrieval persons, so that relatively good targeted property is achieved. The process is capable of automatically screening out objects meeting the conditions without manual input.

Description

A kind of object intelligent search method based on Redis

Technical field

The present invention relates to e-commerce field, be specifically related to a kind of object intelligent search method based on Redis.

Background technology

Along with developing rapidly of computer technology, the level of informatization of human society and growth in the living standard, the degree of dependence of entire society to electronic information platform is also more and more higher.Such as to the retrieval of books, the retrieval to article or the retrieval etc. to product.

Under normal circumstances, when retrieving information, its searcher can only search relative searching object according to a certain feature of searching object, and then result in recall precision reduction, and the result of retrieval is not extensive yet.

The recommendation to object of traditional electronic information platform is usually by the manual typing of operation personnel, and by analyzing the maximum object of the amount of being retrieved, and then commending system is entered in manual typing, finally recommends all searchers again.The result of not carrying out recommending for the demand of different client is like this inaccurate often.

For these reasons, to how, the Search Requirement for searcher self of efficiently and accurately and the recommendation carrying out object more and more come into one's own at present.

Summary of the invention

In view of this, the invention provides a kind of object intelligent search method based on Redis.The Search Requirement of searcher self of efficiently and accurately object can be recommended client.

Based on an object intelligent search method for Redis system, the step realizing the method comprises:

Step one, for a certain object p ₀, definition initial recommendation object sum z; The sort-set data structure of Redis system is utilized to build object p ₀searcher's set A, recommended set Z and other object sets utilize the queue B that the list data structure of Redis system builds;

Step 2, for described object p ₀, gather all searcher of this object in setting-up time section t and retrieval time thereof, if z searcher x _zsearching object p ₀moment be by the information of searcher stored in the object p built ₀in searcher's set A, object p ₀definition key key=object p in searcher's set A ₀, value value=searching object p ₀searcher x _z, weights score=searcher x _zsearching object p ₀moment according to the searching object p of weights score size and searcher ₀the priority moment carry out descending sort, composition searcher set A, is designated as wherein, n is searching object p ₀total number of persons; Searching object p ₀time

Step 3, for m compose initial value be n;

Step 4, from searcher's set A, extract m searcher, extract various combination at every turn, obtain altogether individual Combination nova; By searchers all in each Combination nova to searching object p ₀the retrieval moment be added after carry out descending sort, and insert the tail of the queue of queue B successively;

Step 5, queue B once gone out team's operation, that is: get first combination G in queue B, what obtain each searcher's retrieval in setting-up time section T in this combination G removes object p ₀other objects in addition; By the information of other objects described stored in other object sets built in, other object sets in, definition key key=searcher x _s, value value=searcher x _sother objects of retrieval weights score=retrieves this object time according to weights score size and time carry out descending sort, form other object sets:

$P_{x_s}=\{(P_1^{x_s},T_1^{x_s}),(P_2^{x_s},T_2^{x_s}),\mathrm{...},(P_I^{x_s},T_I^{x_s})\}$

Wherein, I is searcher x _sother object sums retrieved in T time; Retrieval time

$T_1^{x_s}>T_2^{x_s}>\mathrm{...}>T_I^{x_s};$

Step 6, other object sets that each searcher in the combination G taken out in step 5 is retrieved get common factor, judge whether to exist and occur simultaneously, if exist, then perform step 7, otherwise, perform step 8;

Step 7, for each common factor object, the time of this common factor object is retrieved according to each searcher in combination G, obtain the retrieval time the latest of common factor object, and obtain final weights after being added with the searcher quantity m in combination G after converting thereof into timestamp decimal, and by common factor object stored in the recommended set Z built; In recommended set Z, definition key key=object p ₀, value value=common factor object; The final weights of weights score=; According to weights size final in recommended set Z, sort; Obtain the existing object sum in recommended set Z, judge whether to be less than recommended sum z; If so, then step 8 is performed, otherwise, perform step 10;

Step 8, to judge in queue B whether be empty, if so, then perform step 9, otherwise, return step 5;

Step 9, make m from subtracting 1, judge whether m is greater than 0, if so, then returns step 4, otherwise, execution step 10;

Step 10, by z object before in recommended set Z, front z the object that namely weights score is the highest exports as the result for retrieval of recommendation.

Beneficial effect:

The present invention is directed to a certain object to analyze, check all searchers retrieving this object in section sometime, by the analysis to these searchers, be used Redis system Sort-Set and List data structure, sequential sequence is carried out according to the aligning method of the weights given or setting, and then the recommended obtained for different searcher, specific aim is stronger.This process is without the need to manual entry, and Automatic sieve selects eligible object.

Accompanying drawing explanation

Fig. 1 is the process flow diagram of object intelligent recommend method of the present invention.

Embodiment

To develop simultaneously embodiment below in conjunction with accompanying drawing, describe the present invention.

The invention provides a kind of object intelligent search method based on Redis system, basic thought of the present invention is: according to a certain object p ₀all searchers of section sometime, obtain these searchers other searching object information in this time period, the various combination mode obtained according to permutation and combination, obtain and recommend as recommended relevant common factor; In above process, the present invention also utilizes Redis system Sort-Set and List data structure, carries out sequential sequence, thus can obtain the recommended of validity more sometimes according to the weights given or according to the ordering requirements of setting.

Based on an object intelligent search method for Redis system, the step realizing the method comprises:

Step one, for a certain object p ₀, definition initial recommendation object sum z; The sort-set data structure of Redis system is utilized to build object p ₀searcher's set A, recommended set Z and other object sets utilize the queue B that the list data structure of Redis system builds;

Step 2, for described object p ₀, gather all searcher of this object in setting-up time section t and retrieval time thereof, if z searcher x _zsearching object p ₀moment be by the information of searcher stored in the object p built ₀in searcher's set A, wherein definition key key=object p ₀, value value=searching object p ₀searcher x _z, weights score=searcher x _zsearching object p ₀moment according to the searching object p of weights score size and searcher ₀the priority moment carry out descending sort, composition searcher set A, is designated as wherein, n is searching object p ₀total number of persons; Searching object p ₀time $t_{x_1}>t_{x_2}>\mathrm{...}>t_{x_n},$

Step 3, for m compose initial value be n;

Step 4, from searcher's set A, extract m searcher, according to the array mode of planting, extracts various combination at every turn, obtains altogether individual Combination nova.By the mode of permutation and combination, more kinds of retrieval mode can be presented in limited searching object, namely by different array modes, according to the object retrieved of different searcher, find out different common factor objects, and these common factor objects, then recommended exactly.By the object p of searcher's retrievals all in each Combination nova ₀moment be added after carry out descending sort, and insert the tail of the queue of queue B successively; All searcher's searching object p in combining ₀moment be added after, searching object p in this combination can be judged ₀all Time, according to obtain All Time sort, so, when obtaining object common factor in subsequent step, can extract object according to the sequencing of retrieval to occur simultaneously, effectively ensure the ageing of recommended, the accuracy of its recommended can be better.

Step 5, queue B once gone out team's operation, that is: get first combination G in queue B, what obtain each searcher's retrieval in setting-up time section T in this combination G removes object p ₀other objects in addition; By the information of other objects described stored in other object sets built in, wherein, definition key key=searcher x _s, value value=searcher x _sother objects of retrieval weights score=retrieves this object time other retrieved objects are about to according to retrieval time according to weights score size carry out descending sort, form other object sets wherein, I is searcher x _sother object sums retrieved in T time; Retrieval time

Step 6, other object sets that each searcher in the combination G taken out in step 5 is retrieved get common factor, judge whether to exist and occur simultaneously, if exist, then perform step 7, otherwise, perform step 8;

Step 7, for each common factor object, the time of this common factor object is retrieved according to each searcher in combination G, obtain the retrieval time the latest of common factor object, and obtain final weights after being added with the searcher quantity m in combination G after converting thereof into timestamp decimal, and by recommended stored in the recommended set Z built; Wherein, definition key key=object p ₀, value value=common factor object; The final weights of weights score=; According to weights size final in recommended set Z, sort; So, final weights are larger, and it is compound recommendation rules more; Wherein, the integral part of weights is larger, then illustrate that the number of this object retrieval is more; When the integral part of weights is large equally, decimal is larger, then illustrate that retrieval time is newer.Obtain the existing object sum in recommended set Z, judge whether to be less than recommended sum z; If so, then step 8 is performed, otherwise, perform step 10;

Step 8, to judge in queue B whether be empty, if so, then perform step 9, otherwise, return step 5;

Step 9, make m from subtracting 1, judge whether m is greater than 0, if so, then returns step 4, otherwise, execution step 10;

Step 10, by z object before in recommended set Z, front z the object that namely weights score is the highest exports as the result for retrieval of recommendation.

Wherein, when the method being converted to timestamp retrieval time of common factor object is 1970 01 month 01 day 08 Beijing time, 00 point is risen to present total number of seconds for 00 second, afterwards, then will obtain timestamp divided by 10 ¹⁰, and then obtain timestamp decimal.

Such as retrieve commodity, its searching object is then commodity, and searcher is buyer.

In sum, these are only preferred embodiment of the present invention, be not intended to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.

Claims (1)

1. based on an object intelligent search method for Redis system, it is characterized in that, the step realizing the method comprises:

Step one, for a certain object p ₀, definition initial recommendation object sum z; The sort-set data structure of Redis system is utilized to build object p ₀searcher's set A, recommended set Z and other object sets utilize the queue B that the list data structure of Redis system builds;

Step 2, for described object p ₀, gather all searcher of this object in setting-up time section t and retrieval time thereof, if z searcher x _zsearching object p ₀moment be by the information of searcher stored in the object p built ₀in searcher's set A, object p ₀definition key key=object p in searcher's set A ₀, value value=searching object p ₀searcher x _z, weights score=searcher x _zsearching object p ₀moment according to the searching object p of weights score size and searcher ₀the priority moment carry out descending sort, composition searcher set A, is designated as wherein, n is searching object p ₀total number of persons; Searching object p ₀time

Step 3, for m compose initial value be n;

Step 4, from searcher's set A, extract m searcher, extract various combination at every turn, obtain altogether individual Combination nova; By searchers all in each Combination nova to searching object p ₀the retrieval moment be added after carry out descending sort, and insert the tail of the queue of queue B successively;

Step 5, queue B once gone out team's operation, that is: get first combination G in queue B, what obtain each searcher's retrieval in setting-up time section T in this combination G removes object p ₀other objects in addition; By the information of other objects described stored in other object sets built in, other object sets in, definition key key=searcher x _s, value value=searcher x _sother objects of retrieval weights score=retrieves this object time according to weights score size and time carry out descending sort, form other object sets:

$P_{x_s}=\{({p_1}^{x_s},{T_1}^{x_s}),(P_2^{x_s},T_2^{x_s}),...,(P_I^{x_s},T_I^{x_s})\}$

Wherein, I is searcher x _sother object sums retrieved in T time; Retrieval time ${T_1}^{x_s}>T_2^{x_s}>...>T_I^{x_s};$

Step 6, other object sets that each searcher in the combination G taken out in step 5 is retrieved get common factor, judge whether to exist and occur simultaneously, if exist, then perform step 7, otherwise, perform step 8;

Step 7, for each common factor object, the time of this common factor object is retrieved according to each searcher in combination G, obtain the retrieval time the latest of common factor object, and obtain final weights after being added with the searcher quantity m in combination G after converting thereof into timestamp decimal, and by common factor object stored in the recommended set Z built; In recommended set Z, definition key key=object p ₀, value value=common factor object; The final weights of weights score=; According to weights size final in recommended set Z, sort; Obtain the existing object sum in recommended set Z, judge whether to be less than recommended sum z; If so, then step 8 is performed, otherwise, perform step 10;

Step 8, to judge in queue B whether be empty, if so, then perform step 9, otherwise, return step 5;

Step 9, make m from subtracting 1, judge whether m is greater than 0, if so, then returns step 4, otherwise, execution step 10;

Step 10, by z object before in recommended set Z, front z the object that namely weights score is the highest exports as the result for retrieval of recommendation.