CN105718467A - Method and system for evaluating and recommending retrieval algorithms - Google Patents
Method and system for evaluating and recommending retrieval algorithms Download PDFInfo
- Publication number
- CN105718467A CN105718467A CN201410724760.7A CN201410724760A CN105718467A CN 105718467 A CN105718467 A CN 105718467A CN 201410724760 A CN201410724760 A CN 201410724760A CN 105718467 A CN105718467 A CN 105718467A
- Authority
- CN
- China
- Prior art keywords
- list
- record
- searching algorithm
- score
- retrieval
- 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.)
- Granted
Links
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention provides a method and a system for evaluating and recommending retrieval algorithms, and belongs to the technical field of internet. The method comprises the following steps: (S1) determining evaluation conditions according to retrieval background information; (S2) evaluating each retrieval algorithm according to the evaluation conditions; and (S3) obtaining the optimal retrieval algorithm for retrieval calculation and obtaining the retrieval result. According to the method and the system, the optimal retrieval algorithm is obtained through evaluating the retrieval algorithms, so that a user obtains relatively good retrieval experience.
Description
Technical field
The present invention relates to Internet technical field, recommend method and system particularly to a kind of searching algorithm assessment.
Background technology
The standard of the data mining algorithm assessing the Internet at present is based on recall rate and precision or both combinations.Recall rate (RecallRate is also recall ratio) is the ratio of all of relative-record number in the record number retrieved and record storehouse, and measurement is the recall ratio of searching system；Precision is the relative-record number that retrieves and the ratio of record sum retrieved, and measurement is the precision ratio of searching system.Relating to sequence accuracy for the assessment of ordered queue on average to sort the score that on average sorts of score or weighting, this assessment mode is more complicated, inaccurate.
Summary of the invention
For the drawbacks described above of prior art, the technical problem to be solved is how to release optimal searching algorithm, so that user obtains preferably retrieves experience.
For achieving the above object, on the one hand, the present invention provides a kind of searching algorithm assessment recommendation method, and the method comprises the steps:
S1, according to retrieval background information determine evaluation condition；
S2, according to evaluation condition, each searching algorithm is estimated；
S3, acquisition optimal searching algorithm carry out retrieval and calculate, and draw retrieval result.
Preferably, in step s 2, the result ranking in certain time is added up according to search condition, and draw legitimate reading, draw retrieval result respectively according to each retrieval resulting algorithm, the retrieval result of each searching algorithm and legitimate reading are compared, analyze retrieval result and the immediate searching algorithm of legitimate reading.
Preferably, in step s 2, retrieve result according to one and obtain the first list, the second list is obtained according to legitimate reading, first list includes some records, and the second list includes some records, contrasts the first list and the second list, according to the record existed only in the first list, the record existed only in the second list and existence simultaneously and the record in the first list and the second list, analyze retrieval result and the immediate searching algorithm of legitimate reading.
Preferably, in step s 2, described prediction score RS computing formula is:
By the record addition that do not occur in the first list in the second list to the first list, and obtain the 3rd list；
Definition prediction score RS, obtains optimal searching algorithm according to prediction score RS, and described detection score RS computing formula is:
Wherein, k is the record number of the second list；M is the record number of the 3rd list；HjIt is the score of jth record, t in the 3rd listjFor being concurrently present in the score of the record of the first list and the second list；FjThe score of the record of the second list it is not present in for being present in the first list；IjThe score of the record of the second list it is present in for being not present in the first list.
Preferably, described tjBeing multiplied by a coefficient equal to corresponding record at the absolute value of the first list and the difference of secondary series list sorting position, described coefficient confirms according to evaluation condition.
Preferably, described fj=A (m+1), described ij=B (m+1), wherein A, B confirm according to evaluation condition respectively.
On the other hand, the present invention provides a kind of searching algorithm assessment commending system, and this system includes:
Determine module, for determining evaluation condition according to retrieval background information；
Evaluation module, for being estimated each searching algorithm according to evaluation condition；
Recommending module, is used for obtaining optimal searching algorithm and carries out retrieval calculating, and draw retrieval result.
Preferably, described evaluation module, add up the result ranking in certain time according to search condition, and draw legitimate reading, draw retrieval result respectively according to each retrieval resulting algorithm,
Retrieve result according to one and obtain the first list, the second list is obtained according to legitimate reading, first list includes some records, second list includes some records, contrast the first list and the second list, according to the record existed only in the first list, the record existed only in the second list and existence simultaneously and the record in the first list and the second list, analyze retrieval result and the immediate searching algorithm of legitimate reading.
Preferably, described evaluation module, by the record addition that do not occur in the first list in the second list to the first list, and obtain the 3rd list；Definition prediction score RS, obtains optimal searching algorithm according to prediction score RS, and described detection score RS computing formula is:
Wherein, k is the record number of the second list；M is the record number of the 3rd list；HjIt it is the score of jth record in the 3rd list；TjFor being concurrently present in the score of the record of the first list and the second list；FjThe score of the record of the second list it is not present in for being present in the first list；IjThe score of the record of the second list it is present in for being not present in the first list.
Preferably, described tjBeing multiplied by a coefficient equal to corresponding record at the absolute value of the first list and the difference of secondary series list sorting position, described coefficient confirms according to evaluation condition；Described fj=A (m+1), described ij=B (m+1), wherein A, B confirm according to evaluation condition respectively.
The invention provides a kind of to predict the appraisal procedure and system that ordered queue is the mining algorithm of output, the application scenarios of the present invention is usable in the internet arenas such as recommendation and the prediction of assessment commodity, or other have the non-internet field that similar tolerance needs；The technical program quality of the size comprehensive assessment prediction algorithm of one numerical value, and release optimal searching algorithm, so that user obtains preferably retrieves experience；Reduce the visit capacity of server simultaneously.
Accompanying drawing explanation
Fig. 1 is the schematic flow sheet of the searching algorithm assessment recommendation method in embodiments of the invention one；
Fig. 2 is the structural representation of the searching algorithm assessment commending system in the embodiment of the present invention two.
Detailed description of the invention
For making those skilled in the art be more fully understood that technical scheme, below in conjunction with the drawings and specific embodiments, the present invention is described in further detail.
Fig. 1 is the schematic flow sheet of the searching algorithm assessment recommendation method in embodiments of the invention one, as it is shown in figure 1, this searching algorithm assessment recommendation method includes step:
S1, according to retrieval background information determine evaluation condition；
S2, according to evaluation condition, each searching algorithm is estimated；
S3, acquisition optimal searching algorithm carry out retrieval and calculate, and draw retrieval result.
Preferably, in step s 2, the result ranking in certain time is added up according to search condition, and draw legitimate reading, draw retrieval result respectively according to each retrieval resulting algorithm, the retrieval result of each searching algorithm and legitimate reading are compared, analyze retrieval result and the immediate searching algorithm of legitimate reading.
Preferably, in step s 2, retrieve result according to one and obtain the first list, the second list is obtained according to legitimate reading, first list includes some records, and the second list includes some records, contrasts the first list and the second list, according to the record existed only in the first list, the record existed only in the second list and existence simultaneously and the record in the first list and the second list, analyze retrieval result and the immediate searching algorithm of legitimate reading.
Preferably, in step s 2, retrieve result according to one and obtain the first list, the second list is obtained according to legitimate reading, first list includes some records, and the second list includes some records, contrasts the first list and the second list, according to the record existed only in the first list, the record existed only in the second list and existence simultaneously and the record in the first list and the second list, analyze retrieval result and the immediate searching algorithm of legitimate reading.
Preferably, tjBeing multiplied by a coefficient equal to corresponding record at the absolute value of the first list and the difference of secondary series list sorting position, coefficient confirms according to evaluation condition.
Preferably, fj=A (m+1), ij=B (m+1), wherein A, B confirm according to evaluation condition respectively.
Relevant technical staff in the field will be understood that, corresponding with the method for the present invention, the present invention also includes a kind of searching algorithm assessment commending system simultaneously, with said method step correspondingly, as shown in Figure 2, this system comprises determining that module 201, for determining evaluation condition according to retrieval background information；Evaluation module 202, for being estimated each searching algorithm according to evaluation condition；Recommending module 203, is used for obtaining optimal searching algorithm and carries out retrieval calculating, and draw retrieval result.
Preferably, evaluation module 202, add up the result ranking in certain time according to search condition, and draw legitimate reading, add up the result ranking in certain time according to search condition, and draw legitimate reading, draw retrieval result respectively according to each retrieval resulting algorithm,
Retrieve result according to one and obtain the first list, the second list is obtained according to legitimate reading, first list includes some records, second list includes some records, contrast the first list and the second list, according to the record existed only in the first list, the record existed only in the second list and existence simultaneously and the record in the first list and the second list, analyze retrieval result and the immediate searching algorithm of legitimate reading.
Preferably, the record addition that do not occur in the first list in the second list to the first list, and is obtained the 3rd list by evaluation module 202；Definition prediction score RS, obtains optimal searching algorithm according to prediction score RS, and described detection score RS computing formula is:
Wherein, k is the record number of the second list；M is the record number of the 3rd list；HjIt it is the score of jth record in the 3rd list；TjFor being concurrently present in the score of the record of the first list and the second list；FjThe score of the record of the second list it is not present in for being present in the first list；IjThe score of the record of the second list it is present in for being not present in the first list；TjBeing multiplied by a coefficient equal to corresponding record at the absolute value of the first list and the difference of secondary series list sorting position, described coefficient confirms according to evaluation condition；Described fj=A (m+1), described ij=B (m+1), wherein A, B confirm according to evaluation condition respectively.
Wherein, the value of RS is the smaller the better, if predicting the outcome right-on words, the value of RS is 0；K is the number of commodity in real commodity queue；M is all of commodity in prediction queue and true queue；HjBeing the sequence score of jth commodity, point three kinds of situations are as follows: correctly predicted is designated as tj, error prediction be designated as fj, it does not have what be predicted to is designated as ij；TjRepresenting that jth commodity are to simultaneously appear in the correct commodity of prediction in two queues, it is equal to the absolute value of these commodity sorting position in prediction queue with the difference of these commodity sorting position in true queue；FjRepresent that jth commodity are not in true queue, but occur in prediction queue, fj=A (m+1)；IjRepresent that jth commodity are not in prediction queue, but occur in true queue, ij=B (m+1).According to evaluation condition, it is determined that coefficient A, B.
If A=1；B=1；
It is assumed that real commodity queue is T=(B, D, C, E), it was predicted that commodity queue be P=(A, B, C)；For queue T and P, then k=4；M=5 (commodity A, B, C, D, E)；H1=6 (commodity A is misjudged)；H2=1 (commodity B position in P is 2, and the position in T is 1)；H3=0 (commodity C position in T is 3, and position is 3 in P)；H4=6 (commodity D do not have predicted out)；H5=6 (commodity E do not have predicted out)；RS=19/4=4.75.
It is assumed that real commodity queue is T=(D, F, E), it was predicted that commodity queue be P=(E, C, F)；So k=3；M=4 (commodity E, C, F, D)；H1=2 (commodity E position in P is 1, and position is 3 in T)；H2=5 (commodity C is misjudged)；H3=1 (commodity F position in P is 3, and position is 2 in T)；H4=4 (commodity D do not have predicted out)；RS=12/3=4.
Finally, release preferably search method, to provide retrieval to recommend.The application scenarios of the present invention is usable in the internet arenas such as recommendation and the prediction of assessment commodity, or other have the non-internet field that similar tolerance needs.The technical program quality of the size comprehensive assessment prediction algorithm of one numerical value, and release optimal searching algorithm, so that user obtains preferably retrieves experience；Reduce the visit capacity of server simultaneously.
It is understood that the principle that is intended to be merely illustrative of the present of embodiment of above and the illustrative embodiments that adopts, but the invention is not limited in this.For those skilled in the art, without departing from the spirit and substance in the present invention, it is possible to make various modification and improvement, these modification and improvement are also considered as protection scope of the present invention.
Claims (10)
1. a searching algorithm assessment recommendation method, it is characterised in that described method comprises the steps:
S1, according to retrieval background information determine evaluation condition；
S2, according to evaluation condition, each searching algorithm is estimated；
S3, acquisition optimal searching algorithm carry out retrieval and calculate, and draw retrieval result.
2. searching algorithm according to claim 1 assessment recommendation method, it is characterized in that, in step s 2, the result ranking in certain time is added up according to search condition, and draw legitimate reading, draw retrieval result respectively according to each retrieval resulting algorithm, the retrieval result of each searching algorithm and legitimate reading are compared, analyze retrieval result and the immediate searching algorithm of legitimate reading.
3. searching algorithm according to claim 2 assessment recommendation method, it is characterised in that
In step s 2, retrieve result according to one and obtain the first list, the second list is obtained according to legitimate reading, first list includes some records, second list includes some records, contrast the first list and the second list, according to the record existed only in the first list, the record existed only in the second list and existence simultaneously and the record in the first list and the second list, analyze retrieval result and the immediate searching algorithm of legitimate reading.
4. searching algorithm according to claim 3 assessment recommendation method, it is characterised in that in step s 2, described prediction score RS computing formula is:
By the record addition that do not occur in the first list in the second list to the first list, and obtain the 3rd list；
Definition prediction score RS, obtains optimal searching algorithm according to prediction score RS, and described detection score RS computing formula is:
Wherein, k is the record number of the second list；M is the record number of the 3rd list；HjIt is the score of jth record, t in the 3rd listjFor being concurrently present in the score of the record of the first list and the second list；FjThe score of the record of the second list it is not present in for being present in the first list；IjThe score of the record of the second list it is present in for being not present in the first list.
5. searching algorithm according to claim 4 assessment recommendation method, it is characterised in that
Described tjBeing multiplied by a coefficient equal to corresponding record at the absolute value of the first list and the difference of secondary series list sorting position, described coefficient confirms according to evaluation condition.
6. searching algorithm according to claim 4 assessment recommendation method, it is characterised in that
Described fj=A (m+1), described ij=B (m+1), wherein A, B confirm according to evaluation condition respectively.
7. a searching algorithm assessment commending system, it is characterised in that including:
Determine module, for determining evaluation condition according to retrieval background information；
Evaluation module, for being estimated each searching algorithm according to evaluation condition；
Recommending module, is used for obtaining optimal searching algorithm and carries out retrieval calculating, and draw retrieval result.
8. searching algorithm according to claim 7 assessment commending system, it is characterised in that described evaluation module, adds up the result ranking in certain time according to search condition, and draws legitimate reading, draw retrieval result respectively according to each retrieval resulting algorithm,
Retrieve result according to one and obtain the first list, the second list is obtained according to legitimate reading, first list includes some records, second list includes some records, contrast the first list and the second list, according to the record existed only in the first list, the record existed only in the second list and existence simultaneously and the record in the first list and the second list, analyze retrieval result and the immediate searching algorithm of legitimate reading.
9. searching algorithm according to claim 8 assessment commending system, it is characterised in that described evaluation module, by the record addition that do not occur in the first list in the second list to the first list, and obtains the 3rd list；Definition prediction score RS, obtains optimal searching algorithm according to prediction score RS, and described detection score RS computing formula is:
Wherein, k is the record number of the second list；M is the record number of the 3rd list；HjIt it is the score of jth record in the 3rd list；TjFor being concurrently present in the score of the record of the first list and the second list；FjThe score of the record of the second list it is not present in for being present in the first list；IjThe score of the record of the second list it is present in for being not present in the first list.
10. searching algorithm according to claim 8 assessment commending system, it is characterised in that described tjBeing multiplied by a coefficient equal to corresponding record at the absolute value of the first list and the difference of secondary series list sorting position, described coefficient confirms according to evaluation condition；Described fj=A (m+1), described ij=B (m+1), wherein A, B confirm according to evaluation condition respectively.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410724760.7A CN105718467B (en) | 2014-12-03 | 2014-12-03 | Retrieval algorithm evaluation recommendation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410724760.7A CN105718467B (en) | 2014-12-03 | 2014-12-03 | Retrieval algorithm evaluation recommendation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105718467A true CN105718467A (en) | 2016-06-29 |
| CN105718467B CN105718467B (en) | 2020-07-21 |
Family
ID=56145599
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410724760.7A Active CN105718467B (en) | 2014-12-03 | 2014-12-03 | Retrieval algorithm evaluation recommendation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105718467B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109005431A (en) * | 2018-09-18 | 2018-12-14 | 北京腾信创新网络营销技术股份有限公司 | A kind of video evaluations recommender system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102077056A (en) * | 2008-06-27 | 2011-05-25 | 株式会社丰田It开发中心 | Route searching apparatus and route searching method |
| CN102999589A (en) * | 2012-11-15 | 2013-03-27 | Tcl集团股份有限公司 | Method and system for recommending order |
| CN103593411A (en) * | 2013-10-23 | 2014-02-19 | 江苏大学 | Method for testing combination properties of evaluation indexes of search engines and testing device |
| CN103853789A (en) * | 2012-12-06 | 2014-06-11 | 深圳市点行天下科技有限公司 | Method and equipment used for recommending information to user |
| US20140324899A1 (en) * | 2013-04-24 | 2014-10-30 | Kitchology Inc. | Enhanced food information management and presentation on a selective dynamic basis and associated services |
-
2014
- 2014-12-03 CN CN201410724760.7A patent/CN105718467B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102077056A (en) * | 2008-06-27 | 2011-05-25 | 株式会社丰田It开发中心 | Route searching apparatus and route searching method |
| CN102999589A (en) * | 2012-11-15 | 2013-03-27 | Tcl集团股份有限公司 | Method and system for recommending order |
| CN103853789A (en) * | 2012-12-06 | 2014-06-11 | 深圳市点行天下科技有限公司 | Method and equipment used for recommending information to user |
| US20140324899A1 (en) * | 2013-04-24 | 2014-10-30 | Kitchology Inc. | Enhanced food information management and presentation on a selective dynamic basis and associated services |
| CN103593411A (en) * | 2013-10-23 | 2014-02-19 | 江苏大学 | Method for testing combination properties of evaluation indexes of search engines and testing device |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109005431A (en) * | 2018-09-18 | 2018-12-14 | 北京腾信创新网络营销技术股份有限公司 | A kind of video evaluations recommender system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105718467B (en) | 2020-07-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kat et al. | Validation metric based on relative error | |
| Zhou et al. | Remaining useful life prediction of individual units subject to hard failure | |
| Sánchez-González et al. | Quality indicators for business process models from a gateway complexity perspective | |
| Camacho Olmedo et al. | Interest in intermediate soft-classified maps in land change model validation: suitability versus transition potential | |
| RU2017107022A (en) | REMOTE CONTENT DETECTION | |
| Hester et al. | A method for key performance indicator assessment in manufacturing organizations | |
| CN104504583B (en) | The evaluation method of grader | |
| Koc et al. | Comparison of linear correlation and a statistical dependency measure for inferring spatial relation of temperature sensors in buildings | |
| Lu et al. | Interval censored sampling plans for the gamma lifetime model | |
| Parkinson et al. | Evaluating positivist theories of occupant satisfaction: a statistical analysis | |
| You et al. | Optimal exponentially weighted moving average charts with estimated parameters based on median run length and expected median run length | |
| Ding et al. | Ordinal profile monitoring with random explanatory variables | |
| Wang | An imperfect software debugging model considering irregular fluctuation of fault introduction rate | |
| Shirazi et al. | A simulation analysis to study the temporal and spatial aggregations of safety datasets with excess zero observations | |
| CN105786810A (en) | Method and device for establishment of category mapping relation | |
| Gahrooei et al. | An adaptive fused sampling approach of high-accuracy data in the presence of low-accuracy data | |
| Wu | Process performance evaluation based on Taguchi capability index with the consideration of measurement errors | |
| Liu et al. | Integration method for reliability assessment with multi-source incomplete accelerated degradation testing data | |
| Zhang et al. | Non-normal random effects models for immunogenicity assay cut point determination | |
| Sheu et al. | Poisson GWMA control chart | |
| Guo et al. | The variable sampling interval S 2 chart with known or unknown in-control variance | |
| WO2018044955A1 (en) | Systems and methods for measuring collected content significance | |
| Braga et al. | An alternative method for global and partial comparison of two diagnostic systems based on ROC curves | |
| CN105718467A (en) | Method and system for evaluating and recommending retrieval algorithms | |
| Ding et al. | Design of accelerated life test plans under progressive Type II interval censoring with random removals |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20200608 Address after: No. 1-1 Suning Avenue, Xuzhuang Software Park, Xuanwu District, Nanjing City, Jiangsu Province, 210000 Applicant after: Suning Cloud Computing Co.,Ltd. Address before: 210042 Jiangsu, Xuanwu District, Nanjing, Nanjing Road, No. 1, building No. 15 Applicant before: SUNING COMMERCE GROUP Co.,Ltd. |
|
| TA01 | Transfer of patent application right | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20210506 Address after: 518001 unit 3510-130, Luohu business center, 2028 Shennan East Road, Chengdong community, Dongmen street, Luohu District, Shenzhen City, Guangdong Province Patentee after: Shenzhen yunwangwandian Technology Co.,Ltd. Patentee after: NANJING SUNING ELECTRONIC INFORMATION TECHNOLOGY Co.,Ltd. Address before: No.1-1 Suning Avenue, Xuzhuang Software Park, Xuanwu District, Nanjing, Jiangsu Province, 210000 Patentee before: Suning Cloud Computing Co.,Ltd. |
|
| TR01 | Transfer of patent right |