CN109885753A - A kind of method and device for expanding commercial articles searching and recalling - Google Patents
A kind of method and device for expanding commercial articles searching and recalling Download PDFInfo
- Publication number
- CN109885753A CN109885753A CN201910042769.2A CN201910042769A CN109885753A CN 109885753 A CN109885753 A CN 109885753A CN 201910042769 A CN201910042769 A CN 201910042769A CN 109885753 A CN109885753 A CN 109885753A
- Authority
- CN
- China
- Prior art keywords
- word
- commodity
- described search
- search request
- conjunctive
- 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
- 238000000034 method Methods 0.000 title claims abstract description 33
- 230000001419 dependent effect Effects 0.000 claims description 5
- 230000000694 effects Effects 0.000 abstract description 5
- 238000006243 chemical reaction Methods 0.000 abstract description 3
- 241000256844 Apis mellifera Species 0.000 description 5
- 238000011109 contamination Methods 0.000 description 4
- 238000005406 washing Methods 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 239000004615 ingredient Substances 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000007480 spreading Effects 0.000 description 3
- 230000003068 static effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 235000013399 edible fruits Nutrition 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000007689 inspection Methods 0.000 description 2
- 235000001759 Citrus maxima Nutrition 0.000 description 1
- 244000276331 Citrus maxima Species 0.000 description 1
- 241000238557 Decapoda Species 0.000 description 1
- 206010028916 Neologism Diseases 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000011218 segmentation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
Abstract
The invention discloses a kind of method and devices that expansion commercial articles searching is recalled, and method includes: S1: obtaining searching request, match conjunctive word according to the word in searching request;S2: by searching request word and conjunctive word according to default rule be assembled into retrieval unit;S3: matching commodity according to retrieval unit, obtains commodity and recalls result;S4: result is recalled to commodity and is classified, wherein classification includes precisely recalling commodity and obscuring to recall commodity;S5: being divided into accurate show area and fuzzy show area for search show area, shows precisely to recall commodity and obscure respectively and recalls commodity.The present invention is by being that searching request configures different conjunctive words according to preset rules, improve recall precision, suitable for accurately recalling less but the expansion word scene that there are more commodity to recall, it can be under the requirement for guaranteeing correlation, guarantee the bandwagon effect for precisely recalling result, increase the chance for exposure of associated articles, it is whole to improve single-item conversion ratio.
Description
Technical field
The present invention relates to web search technical field, in particular to a kind of method and device for expanding commercial articles searching and recalling.
Background technique
With the popularity of the internet and the rapid development of e-commerce, the application of e-commerce generate increasingly people
Big effect, wherein the most outstanding be, people carry out shopping at network, and shopping at network belongs to network trading, in e-commerce
Occupy extremely important status.Internet user will necessarily first search for when carrying out shopping at network oneself to be wanted purchase or feels emerging
The commodity of interest, and search result is obtained by e-commerce search system, and then screen to the different commodity searched
And comparison, it selects commodity that are wherein most desirable or needing and is bought.
Industry commercial articles searching way is usually to be directed to the inverted indexs of multiple words to find multiple commodity to be detected at present
List obtains the commodity set for meeting all querying conditions in such a way that multichannel seeks common ground.
But for middle-size and small-size e-commerce venture, when commodity library sum is fewer, the shortcomings that above-mentioned way, is:
Meet all querying conditions recalls that result is fewer, but the potential candidate's commodity collection composition and division in a proportion for meeting partial query condition is more.
From the point of view of user, there are certain deviations between psychology intention and searching request, it is expected that eventually finding commodity, it may be necessary to more
Secondary change searching request is experienced bad;From the point of view of businessman, it is meant that candidate commodity chance for exposure is few, influences electric business indirectly
Platform income.
Summary of the invention
In order to solve problems in the prior art, the embodiment of the invention provides a kind of method for recalling of expansion commercial articles searching and
Device, with overcome it is in the prior art meet all querying conditions recall that result is fewer, but meet partial query condition
Potential candidate's commodity collection composition and division in a proportion is more, influences bandwagon effect, causes user experience bad and candidate commodity chance for exposure is few,
Connect the problems such as influencing electric business platform income.
In order to solve the above technical problems, the technical solution adopted by the present invention is that:
On the one hand, a kind of method that expansion commercial articles searching is recalled is provided, described method includes following steps:
S1: obtain searching request, according to described search request in word match conjunctive word;
S2: by described search request in word and the conjunctive word according to default rule be assembled into retrieval unit;
S3: matching commodity according to the retrieval unit, obtains commodity and recalls result;
S4: result is recalled to the commodity and is classified, wherein classification includes precisely recalling commodity and obscuring to recall quotient
Product;
S5: search show area is divided into accurate show area and fuzzy show area, shows precisely recall commodity and mould respectively
Paste recalls commodity.
Further, the step S1 includes:
The synonym and near synonym for obtaining the word in described search request are as conjunctive word;And/or
Classification " key-value " is navigated to dictionary according to foreground, chooses the attribute for belonging to same classified commodity with described search request
As conjunctive word;And/or
By described search request in article word and foreground navigation classification carry out text matches, using sorting key word as pass
Join word;And/or
The label for obtaining dependent merchandise in commodity library is requested according to the label and described search, by relevant commodity
Label is as conjunctive word;And/or
The commodity click logs for obtaining search engine choose conjunctive word according to the co-occurrence rule of described search request.
Further, the step S2 is specifically included:
Judge whether described search request is single word, if so, will be described by the rule of single word assembly retrieval unit
Word and the conjunctive word in searching request are assembled into independent retrieval unit, otherwise, according to the rule of multiple word assembly retrieval units
Then by described search request in word and the conjunctive word be assembled into combined retrieval unit.
Further, the rule of the single word assembly retrieval unit includes:
The upper limit of default conjunctive word quantity obtains the conjunctive word for meeting the upper limit, by described search request in word with
The conjunctive word according to relationship be combined, be assembled into independent retrieval unit.
Further, the rule of the single word assembly retrieval unit further include:
If the conjunctive word be described search request in word synonym and/or near synonym, then the conjunctive word with
Described search request in word and other conjunctive words according to or relationship be combined.
Further, the rule of the multiple word assembly retrieval unit includes:
Judge whether that triggering is recalled according to multiple words that the length of described search request and described search request include, if
Be, first by described search request in each word be assembled into independent retrieval unit according to the rule of single word assembly retrieval unit
Afterwards, then by independent retrieval unit according to relationship be combined, be assembled into combined retrieval unit.
Further, the rule of the multiple word assembly retrieval unit further include:
The multiple words for judging that described search request includes are core word or non-core word, if non-core word, then by institute
State independent retrieval unit that non-core word is assembled into and other independent retrieval units according to or relationship be combined.
Further, the step S3 further include:
If described search request be not single word, when commodity hit be all the conjunctive word when, which should not
It is called back, if described search request is single word, then the commodity can be called back.
Further, the step S4 includes:
When the synonym in the word or the conjunctive word in described search request is hit by commodity, which is divided
Precisely to recall commodity;
When the synonym in the word or the conjunctive word in described search request is not hit by commodity, but described search
When core word in rope request is hit by commodity, which is divided into obscure and recalls commodity;
When the only described conjunctive word is hit by commodity, if described search request is single word, then the commodity are divided
Commodity are recalled to be fuzzy, if described search request is multiple words, are then abandoned the commodity.
On the other hand, a kind of device that expansion commercial articles searching is recalled is provided, described device includes:
Expansion module, for obtaining searching request, according to described search request in word match conjunctive word;
Assembling module, for during described search is requested word and the conjunctive word according to default rule be assembled into inspection
Cable elements;
Module is recalled, for matching commodity according to the retrieval unit, commodity is obtained and recalls result;
Division module is classified for recalling result to the commodity, wherein classification includes precisely recalling commodity and mould
Paste recalls commodity;
Display module is divided into accurate show area and fuzzy show area for that will search for show area, shows precisely call together respectively
It returns commodity and obscures and recall commodity.
Further, the assembling module includes:
Judging unit, for judging whether described search request is single word;
Concatenation unit, for by single word assembly retrieval unit rule by described search request in word and the association
Word be assembled into independent retrieval unit or according to multiple word assembly retrieval units rule by described search request in word and described
Conjunctive word is assembled into combined retrieval unit.
Further, the division module includes:
Analytical unit, for analyze described search request in word and the conjunctive word by commodity hit situation;
Taxon, user according to described search request in word and the conjunctive word by commodity hit situation, by the quotient
Product, which are divided into precisely to recall commodity or obscure, recalls commodity.
Technical solution provided in an embodiment of the present invention has the benefit that
1, the method and device provided by the invention for expanding commercial articles searching and recalling, by being searching request according to preset rules
It configures different conjunctive words, improves recall precision, suitable for accurately recalling less but the expansion word scene that there are more commodity to recall, energy
It is enough to guarantee the bandwagon effect for precisely recalling result under the requirement for guaranteeing correlation, increase the chance for exposure of associated articles, it is whole
Improve single-item conversion ratio;
2, the method and device provided by the invention for expanding commercial articles searching and recalling shows precisely knot by dividing on the page
Fruit and associated articles result, on the one hand, reduce user query number, can quickly find desired commodity, on the other hand, increase
The exposure probability of candidate commodity improves electric business platform income indirectly.
Detailed description of the invention
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those of ordinary skill in the art, without creative efforts, it can also be obtained according to these attached drawings other
Attached drawing.
Fig. 1 is the flow chart for the method that expansion commercial articles searching shown according to an exemplary embodiment is recalled;
Fig. 2 be it is shown according to an exemplary embodiment by single word assembly retrieval unit rule will be in searching request
Word and conjunctive word are assembled into the flow chart of independent retrieval unit;
Fig. 3 be it is shown according to an exemplary embodiment according to multiple word assembly retrieval units rule will be in searching request
Word and conjunctive word be assembled into the flow chart of combined retrieval unit;
Fig. 4 is the apparatus structure schematic diagram that expansion commercial articles searching shown according to an exemplary embodiment is recalled.
Specific embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with attached in the embodiment of the present invention
Figure, technical scheme in the embodiment of the invention is clearly and completely described, it is clear that described embodiment is only this
Invention a part of the embodiment, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art exist
Every other embodiment obtained under the premise of creative work is not made, shall fall within the protection scope of the present invention.
Fig. 1 is the flow chart for the method that expansion commercial articles searching shown according to an exemplary embodiment is recalled, referring to Fig.1 institute
Show, this method comprises the following steps:
S1: obtain searching request, according to described search request in word match conjunctive word.
Specifically, searching request refers to that user is input to the Query in search box, wherein searching request (i.e. Query) can
To be that single word forms, it is also possible to be made of multiple words.After getting the searching request of user, according to every in searching request
A word is associated word matching.What needs to be explained here is that the conjunctive word in the embodiment of the present invention includes the word in searching request
Synonym, near synonym, the expansion word of expansion word and synonym, near synonym one of expansion word etc. or a variety of.
As a kind of preferably embodiment, in the embodiment of the present invention, according to described search request in word matching association
Word includes:
The synonym and near synonym for obtaining the word in described search request are as conjunctive word;And/or
Classification " key-value " is navigated to dictionary according to foreground, chooses the attribute for belonging to same classified commodity with described search request
As conjunctive word;And/or
By described search request in article word and foreground navigation classification carry out text matches, using sorting key word as pass
Join word;And/or
The label for obtaining dependent merchandise in commodity library is requested according to the label and described search, by relevant commodity
Label is as conjunctive word;And/or
The commodity click logs for obtaining search engine choose conjunctive word according to the co-occurrence rule of described search request.
Specifically, in the embodiment of the present invention, when matching conjunctive word, the synonym of the word in preferred searching request and nearly justice
Word.Foreground navigation classification refers to that the navigation point in the navigation bar on the left of electric business platform homepage is interior, the quotient being related to searching request
Product belong to the attribute of same classified commodity as conjunctive word.What needs to be explained here is that selection belongs to the category of same classified commodity
Property as conjunctive word, it is ensured that the commodity that the commodity of Hui Zhao and user are intended to search have correlation, can guarantee it is related
Guarantee user experience while commodity exposure rate.Furthermore it is also possible to by searching request article word and foreground navigation classify into
Row text matches, using sorting key word as conjunctive word.And obtain the label of dependent merchandise in commodity library, according to label and
Searching request, using the label of commodity relevant to the commodity that searching request is related in commodity library as conjunctive word.It needs exist for
It is bright, it further include the commodity click logs obtained in search engine, then according in searching request in the embodiment of the present invention
The co-occurrence rule of word, chooses expansion word as conjunctive word.For example search " apple " obtains this by analyzing commodity click logs
The title of a little single-items obtains " apple " and " black " " white " " 64g " " 128g " " connection " by the methods of similar wordvec
" movement " " customization " " crudely-made articles " etc. there are the possibility of co-occurrence, then from searching request analysing word similitude and rule, next time
When having user to search for " apple 64g ", " apple 128g " can be as preferential match selection;Search for " apple customization " when
It waits, " apple is mobile " can be as preferential match selection.
In addition to above-mentioned several ways, the User action log of search engine can also be obtained, before user session
After input changing rule, determine final searching request.For example, multiple and different searching requests is had input before and after user, according to every
After secondary input, click condition of the user on webpage determines the searching request for finally meeting user's intention.
S2: by described search request in word and the conjunctive word according to default rule be assembled into retrieval unit.
As a kind of preferably embodiment, in the embodiment of the present invention, step S2 is specifically included:
Judge whether described search request is single word, if so, will be described by the rule of single word assembly retrieval unit
Word and the conjunctive word in searching request are assembled into independent retrieval unit, otherwise, according to the rule of multiple word assembly retrieval units
Then by described search request in word and the conjunctive word be assembled into combined retrieval unit.
Specifically, the preset rules of assembled retrieval unit include: that the rule of single word assembly retrieval unit and multiple words are spelled
Fill the rule of retrieval unit.Need to first be determined in assembled retrieval unit the word in searching request quantity be it is single or multiple,
After the quantity for determining the word in searching request, the assembly of retrieval unit is carried out according to corresponding preset rules.
As a kind of preferably embodiment, in the embodiment of the present invention, the rule of single word assembly retrieval unit includes:
The upper limit of default conjunctive word quantity obtains the conjunctive word for meeting the upper limit, by described search request in word with
The conjunctive word according to relationship be combined, be assembled into independent retrieval unit.
Specifically, presetting a upper limit firstly the need of the quantity to conjunctive word.Referring next to shown in Fig. 2, execute with
Lower step:
S1.1: all synonyms are matched as conjunctive word;
S1.2: judging whether the quantity of conjunctive word reaches the upper limit, if so, executing step S1.5, otherwise, executes step
S1.3;
S1.3: near synonym are matched as conjunctive word, and judge whether near synonym matching is completed for the process of conjunctive word, if
It is to then follow the steps S1.4, otherwise return step S1.2;
S1.4: expansion word is matched as conjunctive word, and judges whether expansion word matching is completed for the process of conjunctive word, if
It is to then follow the steps S1.5, otherwise return step S1.2;
S1.5: by searching request word and conjunctive word according to relationship be combined, be assembled into independent retrieval unit.
As a kind of preferably embodiment, in the embodiment of the present invention, the rule of single word assembly retrieval unit further include:
If the conjunctive word be described search request in word synonym and/or near synonym, then the conjunctive word with
Described search request in word and other conjunctive words according to or relationship be combined.
Specifically, assembled independent retrieval unit be default be with relationship, but except some special circumstances.In the present invention
In embodiment, synonym and near synonym and other contamination relationships need to be set as or relationship, can guarantee to close in this way
Join the exposure rate of commodity.For example, the searching request of user's input is " diaper ", synonym includes " paper diaper ", if will
" paper diaper " and other words also according to relationship be combined, then in commodity library some beaten " diaper " or only beaten
The commodity of " paper diaper " label cannot be called back, so that these associated articles cannot be exposed.Therefore, in order to guarantee be associated with quotient
The exposure rate of product, synonym and near synonym and other contamination relationships need to be set as or relationship.
As a kind of preferably embodiment, in the embodiment of the present invention, the rule of multiple word assembly retrieval units includes:
Judge whether that triggering is recalled according to multiple words that the length of described search request and described search request include, if
Be, first by described search request in each word be assembled into independent retrieval unit according to the rule of single word assembly retrieval unit
Afterwards, then by independent retrieval unit according to relationship be combined, be assembled into combined retrieval unit.Specifically, firstly the need of pre-
Different weights first is arranged to different types of word in searching request.With reference to the distribution of the word in title document, calculates and correspond to
Reversed document value idf, this is the static weight of word, referring additionally to the historical search frequency, is calculated for the dynamic of each word
Then state weight combines both into the weight of this word, can finally do for multiple words in current search request (query)
The balance of weight.For example, the static weight of " honeybee " " black " the two words, for this word of honeybee than sparser, static weight is high,
But black searching times are more, and changeable weight is high.If finally search " honeybee black mobile phone ", it is minimum that honeybee eventually becomes weight
, if search " honeybee handbook black " black can become, weight is minimum, and therefore, it is necessary to request current search (query)
In multiple words do the balance of weight.Referring next to shown in Fig. 3, following steps are executed:
S2.1: word segmentation processing is carried out to searching request, obtains the weight and scene information of each word;
S2.2: the ingredient for the multiple words for including according to the length of searching request and searching request judges whether that triggering is recalled,
If so, S2.3 is thened follow the steps, it is no to then follow the steps S2.4;
Specifically, mainly considering whether the problem of neologisms generated to basis participle are searched here.For example " open coffee
You ", its length is not more than 6 bytes, and basis participle uses the result of dictionary matching for " opening coffee/that ", the two words at
Analysis is valueless word, and the document ordering recalled of scattering in this way is possible to disorderly to use on line and actively merge into a word
Method recall document.There are also " kitchen are precious ", " shrimp is sliding " " white shaddock " etc. for similar word.S2.3: every in traversal search request
A word, and judge whether the independent retrieval unit splicing of each word is completed, if so, S2.7 is thened follow the steps, it is no to then follow the steps
S2.5;
S2.4: step S2.7 is executed after selecting other preset query strategies;
S2.5: core word analysis is carried out to multiple words that searching request includes, by non-core word " discarding ";
S2.6: each word in searching request is assembled into independent retrieval list according to the rule of single word assembly retrieval unit
After member, return step S2.3;
S2.7: by independent retrieval unit according to relationship be combined, be assembled into combined retrieval unit.
What needs to be explained here is that " discarding " in step S2.5 is not the meaning of traditional removal, here
Refer in assembled retrieval unit, by non-core word and other contamination relationships be set as or relationship.Non-core word " is lost
Abandoning " needs to consider the ingredient and weight of word, and when the word of ingredient and weight being specified to be less than the weighted average of other words, which just can quilt
It is determined as non-core word.
As a kind of preferably embodiment, in the embodiment of the present invention, the rule of multiple word assembly retrieval units further include:
The multiple words for judging that described search request includes are core word or non-core word, if non-core word, then by institute
State independent retrieval unit that non-core word is assembled into and other independent retrieval units according to or relationship be combined.
Specifically, assembling combination retrieval unit also default be with relationship, but there are also except special circumstances.At this
In inventive embodiments, in order to guarantee the exposure rate of associated articles, the non-core word in searching request is closed with other contaminations
System be set as or relationship.Non-core word includes the word for being intended to not contribute specifically to user's actual search.For example, user inputs
Searching request be " black Haier washing machine ", then include " black ", " Haier ", " washing machine " three in the searching request of user
Word, and currently on the market, there are no black for Haier's washing machine, " black " can be determined as non-core word here.If
According to default rule by " black " and other words according to relationship be combined, will lead to not stamp black in commodity library
Haier's washing machine of label cannot be called back, so that these commodity cannot be exposed.It therefore, equally need to be by non-core word and its
His word be set as or syntagmatic.
S3: matching commodity according to the retrieval unit, obtains commodity and recalls result.
Specifically, the commodity for the condition for meeting retrieval unit (including independent retrieval unit and combined retrieval unit) are called together
It returns, feeds back to user as search result.If searching request is single word, when commodity hit independent retrieval unit, the quotient
Product can be called back, if searching request is not single word, when commodity hit combined retrieval unit, which can be called back.This
In it should be noted that if being that some independent retrieval unit in combined retrieval unit is dropped, then when commodity are hit in addition to this
When other all independent retrieval units of independent retrieval unit, which can be called back.If for example, all words in retrieval unit
Between be all with relationship, then will hit the commodity of all words in retrieval unit and recall.
In addition, it can also include by label from semantically establishing between word and commodity in searching request that commodity, which recall logic,
Mapping, increase latent layer semanteme other than text matches and recall logic, expand recalling for associated articles.
As a kind of preferably embodiment, in the embodiment of the present invention, step S3 further include:
If described search request is not single word (i.e. searching request includes multiple words), what it is when commodity hit is all described
When conjunctive word, which should not be called back, if described search request is single word, then the commodity can be called back.
S4: result is recalled to the commodity and is classified, wherein classification includes precisely recalling commodity and obscuring to recall quotient
Product.
Specifically, when and spreading result less for the accurate result recalled of part searches request is more, by accurate result
It separates with spreading result to user and shows, can be convenient user and check, while working as user's point advanced filter or two minor sorts
When, it can be in disorder to avoid causing to show because of the reason of superset, reduce user experience.
As a kind of preferably embodiment, in the embodiment of the present invention, step S4 includes:
When the synonym in the word or the conjunctive word in described search request is hit by commodity, which is divided
Precisely to recall commodity;
When the synonym in the word or the conjunctive word in described search request is not hit by commodity, but described search
When core word in rope request is hit by commodity, which is divided into obscure and recalls commodity;
When the only described conjunctive word is hit by commodity, if described search request is single word, then the commodity are divided
Commodity are recalled to be fuzzy, if described search request is multiple words, are then abandoned the commodity.
S5: search show area is divided into accurate show area and fuzzy show area, shows precisely recall commodity and mould respectively
Paste recalls commodity.
Specifically, show area is divided into two independent parts, accurate show area and fuzzy show area, between the two may be used
To be distinguished with cut-off rule, when the accurate result recalled is less and spreading result is more, cut-off rule can be played significantly and be mentioned
It is shown as using.
Fig. 4 is the apparatus structure schematic diagram that expansion commercial articles searching shown according to an exemplary embodiment is recalled, referring to Fig. 4
Shown, which includes:
Expansion module, for obtaining searching request, according to described search request in word match conjunctive word.
Specifically, including one or more according to the word matching conjunctive word in searching request:
The synonym and near synonym for obtaining the word in searching request are as conjunctive word;
Classification " key-value " is navigated to dictionary according to foreground, chooses the attribute conduct for belonging to same classified commodity with searching request
Conjunctive word;
By the article word and foreground navigation classification progress text matches in searching request, using sorting key word as association
Word;
The label for obtaining dependent merchandise in commodity library, according to label and searching request, using relevant Commercial goods labels as
Conjunctive word;
The commodity click logs for obtaining search engine choose conjunctive word according to the co-occurrence of searching request rule.
Assembling module, for during described search is requested word and the conjunctive word according to default rule be assembled into inspection
Cable elements.
Module is recalled, for matching commodity according to the retrieval unit, commodity is obtained and recalls result.
Specifically, if searching request is not single word, when commodity hit be all conjunctive word when, which should not be by
It recalls, if searching request is single word, then the commodity can be called back.
Division module is classified for recalling result to the commodity, wherein classification includes precisely recalling commodity and mould
Paste recalls commodity.
Display module is divided into accurate show area and fuzzy show area for that will search for show area, shows precisely call together respectively
It returns commodity and obscures and recall commodity.
As a kind of preferably embodiment, in the embodiment of the present invention, assembling module includes:
Judging unit, for judging whether described search request is single word;
Concatenation unit, for by single word assembly retrieval unit rule by described search request in word and the association
Word be assembled into independent retrieval unit or according to multiple word assembly retrieval units rule by described search request in word and described
Conjunctive word is assembled into combined retrieval unit.
Specifically, the rule of single word assembly retrieval unit includes:
The upper limit of default conjunctive word quantity, obtains the conjunctive word for meeting the upper limit, the word in searching request is pressed with conjunctive word
According to relationship be combined, be assembled into independent retrieval unit.If conjunctive word be the word in searching request synonym and/or
Near synonym, then conjunctive word in searching request word and other conjunctive words according to or relationship be combined.
The rule of multiple word assembly retrieval units includes:
Judge whether that triggering is recalled according to multiple words that the length of searching request and searching request include, if so, will first search
After each word in rope request is assembled into independent retrieval unit according to the rule of single word assembly retrieval unit, then by independent retrieval
Unit according to relationship be combined, be assembled into combined retrieval unit.It should be noted that independent retrieval unit is dressed up group
When closing retrieval unit, also need to judge that multiple words that searching request includes are core word or non-core word, if non-core word, then
By independent retrieval unit that non-core word is assembled into and other independent retrieval units according to or relationship be combined.
As a kind of preferably embodiment, in the embodiment of the present invention, division module includes:
Analytical unit, for analyze described search request in word and the conjunctive word by commodity hit situation;
Taxon, user according to described search request in word and the conjunctive word by commodity hit situation, by the quotient
Product, which are divided into precisely to recall commodity or obscure, recalls commodity.
Specifically, can be divided according to following situations to commodity:
When the synonym in the word or conjunctive word in searching request is hit by commodity, which is divided into and is precisely called together
Return commodity;
When the synonym in the word or conjunctive word in searching request is not hit by commodity, but the core in searching request
When heart word is hit by commodity, which is divided into obscure and recalls commodity;
When only conjunctive word is hit by commodity, if searching request is single word, then the commodity are divided into fuzzy call together
Commodity are returned then to abandon the commodity if searching request is multiple words.
In conclusion technical solution provided in an embodiment of the present invention has the benefit that
1, the method and device provided by the invention for expanding commercial articles searching and recalling, by being searching request according to preset rules
It configures different conjunctive words, improves recall precision, suitable for accurately recalling less but the expansion word scene that there are more commodity to recall, energy
It is enough to guarantee the bandwagon effect for precisely recalling result under the requirement for guaranteeing correlation, increase the chance for exposure of associated articles, it is whole
Improve single-item conversion ratio;
2, the method and device provided by the invention for expanding commercial articles searching and recalling shows precisely knot by dividing on the page
Fruit and associated articles result, on the one hand, reduce user query number, can quickly find desired commodity, on the other hand, increase
The exposure probability of candidate commodity improves electric business platform income indirectly.
It should be understood that the device that expansion commercial articles searching provided by the above embodiment is recalled recalls business in triggering commodity
When, only the example of the division of the above functional modules, in practical application, it can according to need and divide above-mentioned function
With being completed by different functional modules, i.e., the internal structure of device is divided into different functional modules, to complete above description
All or part of function.In addition, provided by the above embodiment expand the device and expand commercial articles searching that commercial articles searching is recalled
The embodiment of the method recalled belongs to same design, and specific implementation process is detailed in embodiment of the method, and which is not described herein again.
Those of ordinary skill in the art will appreciate that realizing that all or part of the steps of above-described embodiment can pass through hardware
It completes, relevant hardware can also be instructed to complete by program, the program can store in a kind of computer-readable
In storage medium, storage medium mentioned above can be read-only memory, disk or CD etc..
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all in spirit of the invention and
Within principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (12)
1. a kind of method for expanding commercial articles searching and recalling, which is characterized in that described method includes following steps:
S1: obtain searching request, according to described search request in word match conjunctive word;
S2: by described search request in word and the conjunctive word according to default rule be assembled into retrieval unit;
S3: matching commodity according to the retrieval unit, obtains commodity and recalls result;
S4: result is recalled to the commodity and is classified, wherein classification includes precisely recalling commodity and obscuring to recall commodity;
S5: being divided into accurate show area and fuzzy show area for search show area, shows precisely to recall commodity and obscure respectively and call together
Return commodity.
2. according to claim 1 expand the method recalled of commercial articles searching, which is characterized in that the step S1 includes:
The synonym and near synonym for obtaining the word in described search request are as conjunctive word;And/or
Classification " key-value " is navigated to dictionary according to foreground, chooses the attribute conduct for belonging to same classified commodity with described search request
Conjunctive word;And/or
By described search request in article word and foreground navigation classification carry out text matches, using sorting key word as be associated with
Word;And/or
The label for obtaining dependent merchandise in commodity library is requested according to the label and described search, by relevant Commercial goods labels
As conjunctive word;And/or
The commodity click logs for obtaining search engine choose conjunctive word according to the co-occurrence rule of described search request.
3. the method according to claim 1 or 2 for expanding commercial articles searching and recalling, which is characterized in that the step S2 is specific
Include:
Judge whether described search request is single word, if so, by the regular by described search of single word assembly retrieval unit
Word and the conjunctive word in request are assembled into independent retrieval unit, otherwise, will according to the rule of multiple word assembly retrieval units
Word and the conjunctive word in described search request are assembled into combined retrieval unit.
4. the method according to claim 3 for expanding commercial articles searching and recalling, which is characterized in that the single word assembly retrieval
The rule of unit includes:
The upper limit of default conjunctive word quantity obtains the conjunctive word for meeting the upper limit, by described search request in word with it is described
Conjunctive word according to relationship be combined, be assembled into independent retrieval unit.
5. the method according to claim 4 for expanding commercial articles searching and recalling, which is characterized in that the single word assembly retrieval
The rule of unit further include:
If the conjunctive word be described search request in word synonym and/or near synonym, then the conjunctive word with it is described
Word and other conjunctive words in searching request according to or relationship be combined.
6. the method according to claim 3 for expanding commercial articles searching and recalling, which is characterized in that the multiple word assembly retrieval
The rule of unit includes:
Judge whether that triggering is recalled according to multiple words that the length of described search request and described search request include, if so, first
By described search request in each word be assembled into independent retrieval unit according to the rule of single word assembly retrieval unit after, then will
Independent retrieval unit according to relationship be combined, be assembled into combined retrieval unit.
7. the method according to claim 6 for expanding commercial articles searching and recalling, which is characterized in that the multiple word assembly retrieval
The rule of unit further include:
The multiple words for judging that described search request includes are core word or non-core word, then will be described non-if non-core word
The independent retrieval unit that core word is assembled into and other independent retrieval units according to or relationship be combined.
8. the method according to claim 3 for expanding commercial articles searching and recalling, which is characterized in that the step S3 further include:
If described search request be not single word, when commodity hit be all the conjunctive word when, which should not be called together
It returns, if described search request is single word, then the commodity can be called back.
9. according to claim 7 expand the method recalled of commercial articles searching, which is characterized in that the step S4 includes:
When the synonym in the word or the conjunctive word in described search request is hit by commodity, which is divided into essence
Standard recalls commodity;
When the synonym in the word or the conjunctive word in described search request is not hit by commodity, but described search is asked
When core word in asking is hit by commodity, which is divided into obscure and recalls commodity;
When the only described conjunctive word is hit by commodity, if described search request is single word, then the commodity are divided into mould
Paste recalls commodity, if described search request is multiple words, then abandons the commodity.
10. a kind of expand the device recalled of commercial articles searching, which is characterized in that described device includes:
Expansion module, for obtaining searching request, according to described search request in word match conjunctive word;
Assembling module, for during described search is requested word and the conjunctive word according to default rule to be assembled into retrieval single
Member;
Module is recalled, for matching commodity according to the retrieval unit, commodity is obtained and recalls result;
Division module is classified for recalling result to the commodity, wherein classification includes precisely recalling commodity and obscuring to call together
Return commodity;
Display module is divided into accurate show area and fuzzy show area for that will search for show area, shows precisely recall quotient respectively
Product recall commodity with fuzzy.
11. the device according to claim 10 for expanding commercial articles searching and recalling, which is characterized in that the assembling module packet
It includes:
Judging unit, for judging whether described search request is single word;
Concatenation unit, for by single word assembly retrieval unit rule by described search request in word and the conjunctive word spell
Dress up independent retrieval unit or according to multiple word assembly retrieval units rule by described search request in word and the association
Word is assembled into combined retrieval unit.
12. the device that expansion commercial articles searching is recalled described in 0 or 11 according to claim 1, which is characterized in that the division module
Include:
Analytical unit, for analyze described search request in word and the conjunctive word by commodity hit situation;
Taxon, user according to described search request in word and the conjunctive word by commodity hit situation, which is drawn
It is divided into precisely recalling commodity or obscuring and recalls commodity.
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910042769.2A CN109885753A (en) | 2019-01-16 | 2019-01-16 | A kind of method and device for expanding commercial articles searching and recalling |
CA3168294A CA3168294A1 (en) | 2019-01-16 | 2019-09-19 | Method of and device for enlarging commodity search recall |
PCT/CN2019/106782 WO2020147332A1 (en) | 2019-01-16 | 2019-09-19 | Method and apparatus for expanding commodity search and recall |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910042769.2A CN109885753A (en) | 2019-01-16 | 2019-01-16 | A kind of method and device for expanding commercial articles searching and recalling |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109885753A true CN109885753A (en) | 2019-06-14 |
Family
ID=66926090
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910042769.2A Pending CN109885753A (en) | 2019-01-16 | 2019-01-16 | A kind of method and device for expanding commercial articles searching and recalling |
Country Status (3)
Country | Link |
---|---|
CN (1) | CN109885753A (en) |
CA (1) | CA3168294A1 (en) |
WO (1) | WO2020147332A1 (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020147332A1 (en) * | 2019-01-16 | 2020-07-23 | 苏宁云计算有限公司 | Method and apparatus for expanding commodity search and recall |
CN111737553A (en) * | 2020-06-16 | 2020-10-02 | 苏州朗动网络科技有限公司 | Method and device for selecting enterprise associated words and storage medium |
CN112507181A (en) * | 2019-09-16 | 2021-03-16 | 百度在线网络技术(北京)有限公司 | Search request classification method and device, electronic equipment and storage medium |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113806519A (en) * | 2021-09-24 | 2021-12-17 | 金蝶软件(中国)有限公司 | Search recall method, device and medium |
CN115878094B (en) * | 2023-01-06 | 2023-07-14 | 粤港澳大湾区数字经济研究院(福田) | Code searching method, device, equipment and storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103207881A (en) * | 2012-01-17 | 2013-07-17 | 阿里巴巴集团控股有限公司 | Query method and unit |
CN103425687A (en) * | 2012-05-21 | 2013-12-04 | 阿里巴巴集团控股有限公司 | Retrieval method and system based on queries |
US9245523B2 (en) * | 2013-07-25 | 2016-01-26 | Nice-Systems Ltd | Method and apparatus for expansion of search queries on large vocabulary continuous speech recognition transcripts |
CN108197098A (en) * | 2017-11-22 | 2018-06-22 | 阿里巴巴集团控股有限公司 | A kind of generation of keyword combined strategy and keyword expansion method, apparatus and equipment |
CN108776901A (en) * | 2018-04-27 | 2018-11-09 | 微梦创科网络科技(中国)有限公司 | Method and system for advertisement recommendation based on search term |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103886093A (en) * | 2014-04-03 | 2014-06-25 | 江苏物联网研究发展中心 | Method for processing synonyms of electronic commerce search engine |
CN106326259A (en) * | 2015-06-26 | 2017-01-11 | 苏宁云商集团股份有限公司 | Construction method and system for commodity labels in search engine, and search method and system |
CN105069086B (en) * | 2015-07-31 | 2017-07-11 | 焦点科技股份有限公司 | A kind of method and system for optimizing ecommerce commercial articles searching |
CN105159884B (en) * | 2015-09-23 | 2018-06-29 | 百度在线网络技术(北京)有限公司 | The method for building up and device of industry dictionary and industry recognition methods and device |
CN109885753A (en) * | 2019-01-16 | 2019-06-14 | 苏宁易购集团股份有限公司 | A kind of method and device for expanding commercial articles searching and recalling |
-
2019
- 2019-01-16 CN CN201910042769.2A patent/CN109885753A/en active Pending
- 2019-09-19 WO PCT/CN2019/106782 patent/WO2020147332A1/en active Application Filing
- 2019-09-19 CA CA3168294A patent/CA3168294A1/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103207881A (en) * | 2012-01-17 | 2013-07-17 | 阿里巴巴集团控股有限公司 | Query method and unit |
CN103425687A (en) * | 2012-05-21 | 2013-12-04 | 阿里巴巴集团控股有限公司 | Retrieval method and system based on queries |
US9245523B2 (en) * | 2013-07-25 | 2016-01-26 | Nice-Systems Ltd | Method and apparatus for expansion of search queries on large vocabulary continuous speech recognition transcripts |
CN108197098A (en) * | 2017-11-22 | 2018-06-22 | 阿里巴巴集团控股有限公司 | A kind of generation of keyword combined strategy and keyword expansion method, apparatus and equipment |
CN108776901A (en) * | 2018-04-27 | 2018-11-09 | 微梦创科网络科技(中国)有限公司 | Method and system for advertisement recommendation based on search term |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2020147332A1 (en) * | 2019-01-16 | 2020-07-23 | 苏宁云计算有限公司 | Method and apparatus for expanding commodity search and recall |
CN112507181A (en) * | 2019-09-16 | 2021-03-16 | 百度在线网络技术(北京)有限公司 | Search request classification method and device, electronic equipment and storage medium |
CN112507181B (en) * | 2019-09-16 | 2023-09-29 | 百度在线网络技术(北京)有限公司 | Search request classification method, device, electronic equipment and storage medium |
CN111737553A (en) * | 2020-06-16 | 2020-10-02 | 苏州朗动网络科技有限公司 | Method and device for selecting enterprise associated words and storage medium |
Also Published As
Publication number | Publication date |
---|---|
WO2020147332A1 (en) | 2020-07-23 |
CA3168294A1 (en) | 2020-07-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109885753A (en) | A kind of method and device for expanding commercial articles searching and recalling | |
US9117006B2 (en) | Recommending keywords | |
TWI615724B (en) | Information push, search method and device based on electronic information-based keyword extraction | |
TWI640878B (en) | Query word fusion method, product information publishing method, search method and system | |
US8650198B2 (en) | Systems and methods for facilitating the gathering of open source intelligence | |
CN111260437B (en) | Product recommendation method based on commodity-aspect-level emotion mining and fuzzy decision | |
CN106339502A (en) | Modeling recommendation method based on user behavior data fragmentation cluster | |
CN111695022B (en) | Interest searching method based on knowledge graph visualization | |
CN103020066B (en) | A kind of method and apparatus identifying search need | |
WO2015055094A1 (en) | Method and device for providing screening conditions and method and device for searching | |
CN105723402A (en) | Systems and methods for determining influencers in a social data network | |
CN104462336A (en) | Information pushing method and device | |
CN103365904B (en) | A kind of advertising message searching method and system | |
CN110532351B (en) | Recommendation word display method, device and equipment and computer readable storage medium | |
CN108280124B (en) | Product classification method and device, ranking list generation method and device, and electronic equipment | |
CN107256513A (en) | Method and device is recommended in a kind of collocation of object | |
CN107203520A (en) | The method for building up of hotel's sentiment dictionary, the sentiment analysis method and system of comment | |
CN108228556A (en) | Key phrase extracting method and device | |
CN111444304A (en) | Search ranking method and device | |
CN105931082B (en) | Commodity category keyword extraction method and device | |
CN112579854A (en) | Information processing method, device, equipment and storage medium | |
CN111428007B (en) | Cross-platform based synchronous push feedback method | |
CN108153781A (en) | The method and apparatus for extracting the keyword of business scope | |
CN108205545B (en) | Method and equipment for providing recommendation information for user | |
KR101902460B1 (en) | Device for document categorizing |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190614 |