CN105303404A - Method for fast recognition of user interest points - Google Patents
Method for fast recognition of user interest points Download PDFInfo
- Publication number
- CN105303404A CN105303404A CN201510702141.2A CN201510702141A CN105303404A CN 105303404 A CN105303404 A CN 105303404A CN 201510702141 A CN201510702141 A CN 201510702141A CN 105303404 A CN105303404 A CN 105303404A
- Authority
- CN
- China
- Prior art keywords
- word
- evaluation
- picture
- user interest
- density
- 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
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention discloses a method for fast recognition of user interest points. The method provided by the invention comprises: establishing an on-line detection/recording mechanism based on a Web browser, and realizing a character and picture show detection area based on the Web browser; allowing users who participate in the detection to access the characters and the pictures in the show detection area through the Internet on-line Web browser and mark feeling results through a mouse after reading contents; recording mark results to a background in a unified manner; calculating evaluation density of all the mark results on each minimum rating unit of the characters and the pictures, and performing visualization display in the research and analysis interface of the background according to the evaluation densities; and selecting different screening threshold values, and obtaining analysis results of unified user interest points. The method for fast recognition of user interest points is capable of analyzing and discovering users' concerns in the user behavior data, and has simple method, low cost, fast speed and accurate and deep analysis.
Description
Technical field
The present invention relates to internet arena, specifically a kind of method of quick identification user interest point.
Background technology
1. product focus before measurement
In Business Studies, before the launch of domestic consumer, need to do a series of product before measurement.One of them important test detects the impression of user to product advertising publicity characteristic.Producer/researchist and advertising creative personnel, wish to find that user is to the place paid close attention to most in products propaganda characteristic (like with dislike) by before measurement, thus improved, the point of interest that outstanding user likes in actual advertisement, weaken the problem of dislike, so that the more popular acceptance of product and purchase.
The publicity characteristic of test, mainly two kinds of contents: slogan (for copy/broadcast/video ads with sheet voice) and publicity pictures (for static advertising occasions such as the display advertisings of outdoor advertising/online).Slogan, by the word less than 100 words, describe the product performance needing to transmit, and advertising map is the visual presentation to product performance.
Above content combines generation by produce sector of producer with advertising creative design department usually.In focus before measurement, intuitively passed judgment on by the targeted customer of reality, researchist and creative personnel can revise original publicity characteristic for evaluation result, strengthen the point of interest of forward, delete the publicity point of negative/shortage transmission capacity.After correlated process completes, related content is content distributed as the product advertising of reality.
2. flow gauge and method before focus
In Business Studies, have an informal discussion under relevant before measurement research extensively adopts line and detect the pattern with manual analysis.The some users of each invitation (number 30-100 is indefinite, determines according to product), to Test Field, are divided into many groups (5-10 groups), often organize 8-10 people.Often organize people and enter test site (meeting room of having an informal discussion under being generally line), carry out the focus before measurement of being correlated with.Scene has related researcher to exchange, and responsible coordination is answerred questions and logging test results.
The concrete grammar of test is as follows:
(1) content that user provides test is surveyed to each ginseng.Respective organized formats is as follows:
Publicity text: be printed on normal paper.Slogan divides by the characteristic of sentence/detection, and each sentence/characteristic that will evaluate gives a particular number.
Advertising map: a large photo or placard, is divided into zones of different by the point of design, and is numbered.
(2) ginseng surveys user, makes following judgement according to the impression of self:
Publicity text: according to the impression to publicity literary composition, after the numbering of specifying sentence/characteristic, writes the grading evaluation (generally including, neutral, poor) of oneself
Advertising map: according to the visual sense to plate, at different evaluation districts circle note, writes the grading evaluation of oneself.
(3) result marked, as test result, studied personnel collect and record.Some other feedbacks of researcher's on-the-spot meeting record.
After all test results complete, researchist, according to the annotation results of all words and pattern, by manually checking and summing up, concludes out important focus, and as final before measurement result.
3. some problems of existing before measurement method
Point of interest before measurement research method presented hereinbefore, employs for many years in research industry.Along with current user interest and the fast-changing commercial trends of product, expose increasing problem.
The pattern of testing under subscribers feeder, consuming time, cost is high, and test result applicability exists risk.Because user will carry out test site, take time and effort, invite cost very large (invite per capita and take in hundreds of unit), complete whole test and usually want more than one week.Due to cost and time restriction, a small amount of user (being no more than 100) can only be tested, such test subscriber carrys out source distribution poor (comprehensive not), the usual place that can test also limited (minority big city), user's impression in most of area is left in the basket, the applicability of final influence research result.For the product of national sale/advertisement, just there is great risk.
To the detection of user interest point, granularity is too thick, often omits valuable point.In reality detects, for the ease of implementing, predefined is wanted to the point detected, and granularity is thicker.If publicity literature is all divide by sentence in advance, user marks whole sentence, can only pass judgment on the quality of each sentence like this.And user actual often have the different piece in sentence to experience (such as certain word is with deep impression above, and certain phrase of later half sentence is dull) more.For picture, usually divide and mark by key area, but user's vision intuitively to put be often random, focus is across multizone sometimes, sometimes belongs to prior undefined region.These actual/random evaluation points produced, the impression that more representative of consumer is deep in fact, more valuable, but cannot effectively support, so be often rejected due to the existing mark analytical approach of researcher.
To the overall study of testing result, based on artificial experience, efficiency is low, lacks and quantizes unified standard, and result fluctuation is large.Usually manually checked the evaluation result analyzing up to a hundred words and picture by researcher, now researcher usually only pays close attention to it and thinks most important content, then analyses in depth.And relevant perception and selection, be the personal experience based on researcher and ability completely, there is no unified evaluation criterion and quantitative evaluation result, cause the fluctuation of result of study larger.
Summary of the invention
The object of the present invention is to provide a kind of method that method is simple, cost is low, analysis identifies user interest point accurately fast, to solve the problem proposed in above-mentioned background technology.
For achieving the above object, the invention provides following technical scheme:
A method for quick identification user interest point, comprises the following steps:
1) set up the on-line checkingi/recording mechanism of sing on web browser, what achieve the word of sing on web browser and picture represents detection zone; User, by Web browser, the word of display marks arbitrarily and evaluates; The arbitrary region of the picture of display is arbitrarily marked and evaluated;
2) participate in the user detected, accessed the word and picture that represent in detection zone by the online Web browser in internet, after reading content, experience result by mouse mark; Annotation results is unified is recorded to backstage;
3) for the annotation results of all records in backstage, the evaluation density of all annotation results on each minimum Ranking Unit of calculating word/picture, represent with the coverage rate of each minimum Ranking Unit, wherein the minimum Ranking Unit of word is word, and the minimum Ranking Unit region of picture is minimum is pixel; And according to evaluation density, carry out visual presentation researching and analysing in interface of backstage;
4) according to the evaluation density calculated and visual presentation, select different screening threshold values, just can obtain the analysis result of unified user interest point.
As the further scheme of the present invention: mark online the point of interest of the picture arbitrary region of display.
As the further scheme of the present invention:, be accurate to word to the online arbitrarily mark support of word, the record attribute of annotation results comprises initial sum termination locations and the rating information of the word of mark.
As the further scheme of the present invention: the annotation results of word/picture comprises tab area and rating information.
As the further scheme of the present invention: when carrying out visual presentation, the evaluation density of word changes by color and font and presents, and the evaluation density of picture presents by thermodynamic chart mode.
As the further scheme of the present invention: step 3) the middle computation process evaluating density, as described below:
The set of hypothesis evaluation rank value be E (E1, E2 ... Ei ... En), Ei is an Assessment for classification value; The set in all minimum Ranking Unit regions be A (A1, A2 ... Am ... An), in character area, Am is the minimum Ranking Unit region of each word, is the minimum Ranking Unit region of single pixel or the set of rectangular pixels point in picture region; The evaluation of estimate of each user u to minimum Ranking Unit region Am is E (u, m, i), then the Assessment for classification value of minimum Ranking Unit region Am is the evaluation frequency F (m of Ei, i) be: F (m, i)=sum (E (u, m, i)), wherein u is the set of all users; Evaluating density D (m, i) is: D (m, i)=f (F (m, i)), and f () function is the density calculation function evaluating frequency result.
Compared with prior art, the invention has the beneficial effects as follows:
The present invention can analyze and find the focus of user from user behavior data, and method is simple, and complexity is low, test data of experiment achieves cost is low, speed fast, analyze accurate and deep advantage, and result no longer relies on researcher profile.This illustrates that the present invention has good adaptability, has objective, reliable, comprehensive feature, has a good application prospect.
Embodiment
Below in conjunction with the embodiment of the present invention, be clearly and completely described the technical scheme in the embodiment of the present invention, obviously, described embodiment is only the present invention's part embodiment, instead of whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art, not making the every other embodiment obtained under creative work prerequisite, belong to the scope of protection of the invention.
Embodiment 1
In the embodiment of the present invention, be in existing user study about the deficiency of the point of interest analytical approach of product before measurement, propose to detect and quantitative analysis method based on online random point of interest.The method, based on on-line checkingi pattern, can gather the completely autonomous evaluation region of user and content by technological means, and support the mark records of minimum particle size.When analyzing by the quantification measure algorithm based on evaluation density, the assessment result that the standard that obtains is unified.
The present invention completes on computers, after user selects relevant range and mark by Web browser, successively containing following steps:
Step 1 quantizes user annotation result
User annotation result is the basis of follow-up characteristic.The interface annotation results of user's sing on web browser, can not directly use, and quantize to be converted into the user annotation result that can be used for analyzing.Relevant work comprises:
A label character: to the word that browser is selected, by word (not according to sentence) for unit carries out the sequential counting of position and the quantification of the overall situation.The quantified property of annotation results is as shown in table 1.
Table 1
B image labeling: to the selection of user on browser to certain region of picture is according to pixels that unit quantizes with scaling.Because browser limits, the actual picture shown may vary in size (certain proportional zoom) with former figure, so the image-region of the browser of user's actual selection, the pixel (calculating with the relativity shift in the browser upper left corner) of its coordinate points and length, all to map according to scaling respective opposed and go back, obtain the official portrait vegetarian refreshments position that former figure is corresponding.The quantified property of such picture annotation results is as shown in table 2.
Table 2
Step 2. calculates the evaluation density of each Basic Evaluation unit area
Based on all annotation results, by each Basic Evaluation unit area, calculate corresponding assessment grade and relevant covering.The minimum Ranking Unit of word is word, and the minimum Ranking Unit of picture is pixel.Calculate respectively:
The set of hypothesis evaluation rank value (as: good, in, poor) be E (E1, E2 ... Ei ... En), Ei is an Assessment for classification value.And the set in all minimum Ranking Unit regions is A (A1, A2, Am ... An), in character area, Am is the minimum Ranking Unit region (containing correlation name and deviation post) of each word, is the minimum Ranking Unit region (coordinated manner) of single pixel or the set of rectangular pixels point in picture region.The evaluation of estimate of each user u to minimum Ranking Unit region Am is E (u, m, i) (actual is 1 or 0), to be then evaluation frequency F (m, i) of Ei be the Assessment for classification value in Am region:
F (m, i)=sum (E (u, m, i)), (u is the set of all evaluation users)
Evaluating density D (m, i) is D (m, i)=f (F (m, i)), and f () function is the density calculation function (can use different function according to disparity items) evaluating frequency result.
To each Basic Evaluation unit, according to previous methods, calculate the evaluation density of all Assessment for classifications.
Now, the quantification assessment result in all evaluation points and region has been obtained.
Step 3 polymerization density region, the user interest point that visual analyzing is final
For analysis result, according to Assessment for classification value, to evaluating the region of density higher than certain screening threshold value, carry out being polymerized the main points as analyzing.In order to better practical operation, according to evaluation density, different display modes is set to raw content, effect can be found out intuitively.Method is as follows:
For word content, the color of each word corresponding with Assessment for classification (can by green to red), and the size of word, evaluate density values corresponding (in order to better showing interface, usually will do the smoothing processing of value, as sqrt) with corresponding.Such word content is exactly be made up of the word that a pile continuous print size is different with color, and the point that user pays close attention to is made up of larger word, and the rank of evaluation is relevant to color.
For image content, the covering Show Color of each pixel/unit area corresponding with evaluation density (can by green to red by degree), evaluates density showing by original color lower than certain threshold value.Points a large amount of in such picture cover by thermodynamic chart pattern, according to color change to color from red to green, represent the change of user to the concern in individual region.
From above display result, final user interest point region (usually getting a most important 3-5 point of interest) can be obtained by Direct Analysis.
In order to verify validity and the versatility of the inventive method, carry out correlation test.
From the test products of our cooperation manufacturer, the product tested under selecting once to do actual line: certain coffee (copy test) and toothbrush (display advertising is tested), Related product forum correspondence has respectively recruited 100 volunteers from network, and the line contrasted is tested.
Result is as follows:
Test speed and cost: under line, test test subscriber 30 every day, the average > consuming time of each user 2 hours (comprise traffic and grouping is waited for), all users are from Shanghai, and 100 user tests completed with 3 days, per capita expense 200 yuan.Line is tested, and each user completes test and only needs 5 minutes, and from the whole nation, an afternoon has only been used in 100 user tests, per capita expense 20 yuan.
The granularity of test: in traditional test, content of text marks with simple sentence, and when new line marks, user's reality does specific evaluation to the specific word in text, across the phrase etc. of sentence in a large number; During figure built-in testing, the tab area of user on line, major part is not exclusively identical with the predefine region that marks under line, there is the point of a lot of refinement.
Can find from the final analysis result of coffee copy, there is multiple different font (representing different concern interest-degrees), and mostly focus be word or phrase, not whole sentence.In classic method, the label frequency arrangement of simple sentence pressed in word, and on-line analysis method accurately can find two prior words or phrase in simple sentence.From toothbrush display advertising user interest selection result in.For the ease of checking, being Basic Evaluation unit area based on 20*20 pixel, based on evaluation density, some regions of low evaluation density being screened out, remaining region being carried out thermodynamic chart calculating (central area color is dark).Can find toothbrush head, bar portion, handle upper portion and bottom, the impression that user has nothing in common with each other, even there is difference handle top and the bottom.In picture, occurred a large amount of thermodynamic chart focus, the position of these focus, the region more than classic method is wanted accurately.
To those skilled in the art, obviously the invention is not restricted to the details of above-mentioned one exemplary embodiment, and when not deviating from spirit of the present invention or essential characteristic, the present invention can be realized in other specific forms.Therefore, no matter from which point, all should embodiment be regarded as exemplary, and be nonrestrictive, scope of the present invention is limited by claims instead of above-mentioned explanation, and all changes be therefore intended in the implication of the equivalency by dropping on claim and scope are included in the present invention.
In addition, be to be understood that, although this instructions is described according to embodiment, but not each embodiment only comprises an independently technical scheme, this narrating mode of instructions is only for clarity sake, those skilled in the art should by instructions integrally, and the technical scheme in each embodiment also through appropriately combined, can form other embodiments that it will be appreciated by those skilled in the art that.
Claims (6)
1. identify a method for user interest point fast, it is characterized in that, comprise the following steps:
1) set up the on-line checkingi/recording mechanism of sing on web browser, what achieve the word of sing on web browser and picture represents detection zone; User, by Web browser, the word of display marks arbitrarily and evaluates; The arbitrary region of the picture of display is arbitrarily marked and evaluated;
2) participate in the user detected, accessed the word and picture that represent in detection zone by the online Web browser in internet, after reading content, experience result by mouse mark; Annotation results is unified is recorded to backstage;
3) for the annotation results of all records in backstage, the evaluation density of all annotation results on each minimum Ranking Unit of calculating word/picture, represent with the coverage rate of each minimum Ranking Unit, wherein the minimum Ranking Unit of word is word, and the minimum Ranking Unit region of picture is minimum is pixel; And according to evaluation density, carry out visual presentation researching and analysing in interface of backstage;
4) according to the evaluation density calculated and visual presentation, select different screening threshold values, just can obtain the analysis result of unified user interest point.
2. the method for quick identification user interest point according to claim 1, is characterized in that, marks online the point of interest of the picture arbitrary region of display.
3. the method for quick identification user interest point according to claim 1, is characterized in that, be accurate to word to the online arbitrarily mark support of word, the record attribute of annotation results comprises initial sum termination locations and the rating information of the word of mark.
4. the method for quick identification user interest point according to claim 1, is characterized in that, the annotation results of word/picture comprises tab area and rating information.
5. the method for quick identification user interest point according to claim 1, is characterized in that, when carrying out visual presentation, the evaluation density of word changes by color and font and presents, and the evaluation density of picture presents by thermodynamic chart mode.
6. the method for quick identification user interest point according to claim 1, is characterized in that, step 3) the middle computation process evaluating density, as described below:
The set of hypothesis evaluation rank value be E (E1, E2 ... Ei ... En), Ei is an Assessment for classification value; The set in all minimum Ranking Unit regions be A (A1, A2 ... Am ... An), in character area, Am is the minimum Ranking Unit region of each word, is the minimum Ranking Unit region of single pixel or the set of rectangular pixels point in picture region; The evaluation of estimate of each user u to minimum Ranking Unit region Am is E (u, m, i), then the Assessment for classification value of minimum Ranking Unit region Am is the evaluation frequency F (m of Ei, i) be: F (m, i)=sum (E (u, m, i)), wherein u is the set of all users; Evaluating density D (m, i) is: D (m, i)=f (F (m, i)), and f () function is the density calculation function evaluating frequency result.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510702141.2A CN105303404A (en) | 2015-10-23 | 2015-10-23 | Method for fast recognition of user interest points |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510702141.2A CN105303404A (en) | 2015-10-23 | 2015-10-23 | Method for fast recognition of user interest points |
Publications (1)
Publication Number | Publication Date |
---|---|
CN105303404A true CN105303404A (en) | 2016-02-03 |
Family
ID=55200637
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510702141.2A Pending CN105303404A (en) | 2015-10-23 | 2015-10-23 | Method for fast recognition of user interest points |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105303404A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110083792A (en) * | 2019-05-06 | 2019-08-02 | 武汉微麦互联网络科技有限公司 | A kind of file management method based on wechat public service platform |
CN112712224A (en) * | 2019-10-25 | 2021-04-27 | 华晨宝马汽车有限公司 | Customer-oriented product quality management method, system, device, and medium |
CN112989020A (en) * | 2019-12-17 | 2021-06-18 | 北京沃东天骏信息技术有限公司 | Information processing method, apparatus and computer readable storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102402593A (en) * | 2010-11-05 | 2012-04-04 | 微软公司 | Multi-modal approach to search query input |
CN103345465A (en) * | 2013-06-28 | 2013-10-09 | 宇龙计算机通信科技(深圳)有限公司 | Method and device for labeling and displaying multi-media files |
CN103582898A (en) * | 2011-05-26 | 2014-02-12 | 创慧株式会社 | Market evaluation investigation device and investigation method |
US20140184866A1 (en) * | 2012-12-28 | 2014-07-03 | Canon Kabushiki Kaisha | Image pickup element, image pickup apparatus, and method and program for controlling the same |
CN104680393A (en) * | 2013-12-02 | 2015-06-03 | 章文贤 | Interactive advertisement method based on image contents and matching |
-
2015
- 2015-10-23 CN CN201510702141.2A patent/CN105303404A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102402593A (en) * | 2010-11-05 | 2012-04-04 | 微软公司 | Multi-modal approach to search query input |
CN103582898A (en) * | 2011-05-26 | 2014-02-12 | 创慧株式会社 | Market evaluation investigation device and investigation method |
US20140184866A1 (en) * | 2012-12-28 | 2014-07-03 | Canon Kabushiki Kaisha | Image pickup element, image pickup apparatus, and method and program for controlling the same |
CN103345465A (en) * | 2013-06-28 | 2013-10-09 | 宇龙计算机通信科技(深圳)有限公司 | Method and device for labeling and displaying multi-media files |
CN104680393A (en) * | 2013-12-02 | 2015-06-03 | 章文贤 | Interactive advertisement method based on image contents and matching |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110083792A (en) * | 2019-05-06 | 2019-08-02 | 武汉微麦互联网络科技有限公司 | A kind of file management method based on wechat public service platform |
CN112712224A (en) * | 2019-10-25 | 2021-04-27 | 华晨宝马汽车有限公司 | Customer-oriented product quality management method, system, device, and medium |
CN112989020A (en) * | 2019-12-17 | 2021-06-18 | 北京沃东天骏信息技术有限公司 | Information processing method, apparatus and computer readable storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Khalayleh et al. | The impact of digital content of marketing mix on marketing performance: An experimental study at five-star hotels in Jordan | |
CN106651424B (en) | Power user portrait establishing and analyzing method based on big data technology | |
Acosta et al. | Capital accumulation, trade liberalization, and rising wage inequality: the case of Argentina | |
Dzyabura et al. | Leveraging the power of images in managing product return rates | |
Khaliq et al. | Leadership and decision making among SMEs: management accounting information and the moderating role of cloud computing | |
Becker et al. | Wrapping up the Europe 2020 strategy: A multidimensional indicator analysis | |
Shayaa et al. | Linking consumer confidence index and social media sentiment analysis | |
CN115730858B (en) | Cultural product design intelligent management platform based on big data | |
Jinling et al. | Modeling E-commerce website quality with quality function deployment | |
Israr | Training effectiveness and performance of micro, small and medium sized enterprises | |
CN105303404A (en) | Method for fast recognition of user interest points | |
Hong et al. | Measuring product type with dynamics of online product review variance | |
Fan et al. | Visual complexity of chinese ink paintings | |
Alcalde et al. | Framing bias: The effect of figure presentation on seismic interpretation | |
Widayat et al. | Bibliometric analysis and visualization articles on presidential election in social media indexed in Scopus by Indonesian authors | |
CN116051281A (en) | Method and system for conducting goods matching transaction on goods and advertisement medium point positions | |
CN117575526A (en) | Image design visual conveying system based on big data | |
Antunes et al. | Who are the tourists sharing content on social media? Behaviour and characteristics | |
Menegaki et al. | Assessing the chromatic contrast in open surface excavations: a comparative study between subjective and quantitative approaches | |
Indayani | Calculation of depreciation Fixed Assets based on Government Accounting Standards and their impact on Financial Statements | |
Mayr et al. | Visualizing Biographical Trajectories by Historical Artifacts: A Case Study Based on the Photography Collection of Charles W. Cushman. | |
Huang et al. | Harnessing machine learning for landscape character management in a shallow relief region of China | |
CN115601081A (en) | Online digital exhibition service management system based on big data analysis | |
JP2007304842A (en) | Community value calculation method and community value calculation system calculation device | |
CN107451898A (en) | A kind of luxury goods rent method and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160203 |
|
WD01 | Invention patent application deemed withdrawn after publication |