際際滷

際際滷Share a Scribd company logo

Behavioral targeting

Download as PPTX, PDF
Avelin Huo
Avelin Huo

- Behavioral targeting (BT) uses an individual's web browsing history to select advertisements. The document evaluates how much BT can help online advertising. - The experiment finds users who clicked the same ad are over 90 times more similar than those who clicked different ads, supporting the assumption of BT. BT improved ad click-through rates by up to 670% on average through user segmentation. - Tracking short-term search behavior performed better for BT than long-term browsing behavior. Overall, the results indicate BT has significant potential to improve online advertising performance.

1 of 32
How much can Behavioral Targeting Help Online Advertising?Jun Yan, Ning Liu, Gang Wang, Wen Zhang, Yun Jiang, Zheng Chen, WWW 2009Advisor: Chia-Hui ChangStudent: Kuan-Hua HuoDate: 2010-09-071
OutlineIntroductionBehavioral TargetingExperiment Configuration BT ResultsConclusions2
InstructionBehavioral Targeting (BT)BT uses information collected on an individual's web-browsing behavior, such as the pages they have visited or the searches they have made, to select which advertisements to display to that individual. Practitioners believe this helps them deliver their online advertisements to the users who are most likely to be influenced by them.3
How much can BT help online advertising?Does BT truly have the ability to help online advertising?How much can BT help online advertising using commonly used evaluation metrics?What BT strategy works better than others for ads delivery?4
Does BT truly have the ability to help online advertising?Users have similar search or browsing behaviorssimilar interestshigher probability to click the same adOnline users can be grouped into different user segments according to their behaviorswithin- and between- ads user similaritiesthe users who clicked the same ad can be over 90 times more similar than the users who clicked different ads.5
How much can BT help online advertising using commonly used evaluation metrics?We propose to observe how much BT can improve ads CTR through the segmentation of users into a number of small groups for targeted ads delivery.Through studying ads CTR before and after user segmentation for ads deliveryads CTR can be improved by as much as 670% over all the ads we collected.6
What BT strategy works better than others for ads delivery?According to the definition of Behavioral Targeting (BT), there are two strategies to represent the users behaviorWeb browsing behavior => users clicked pagesSearch behavior => search querieswe draw the conclusion that user search queries can perform several times better than user browsing behaviortracking the short term user behaviors are more effective than tracking the long term user behaviors7
Behavioral TargetingTo represent user behavior by their page-viewsWe consider the clicked URLs of search users as their profiles All the users can be considered as a user-by-URL matrixThe classical Term Frequency Inverse Document Frequency (TFIDF) indexingeach user as a document and each URL as a term8
Behavioral Targeting cont.Mathematically, all users are represented by a real valued matrixg is the total number of users l is the total number of URLs A user is a row of U, a real valued vector with the weight for each entry to be, 9
Behavioral Targeting cont.We build the user behavioral profile by all terms that appear in a users queries as his previous behaviors. We can represent each user in the Bag of Words (BOW) model each term is considered as a feature.each user is represented by BOW with corresponding term frequencyWe use the same TFIDF indexing10
Behavioral Targeting cont.Two different BT strategiesThe long term user behavior (7 days user behavior) The short term user behavior (1 days user behavior)Validate and compare four different BT Strategies1. LP: using Long term user behavior all through the seven days and representing the user behavior by Page-views; 2. LQ: using Long term user behavior all through the seven days and representing the user behavior by Query terms; 3. SP: using Short term user behavior (1 day) and representing user behavior by Page-views; 4. SQ: using Short term user behavior (1 day) and representing user behavior by Query terms.11
Experiment ConfigurationSymbols and Experiment Setup which will be used throughout the experimentswith detailed experiment configurationsEvaluation Metrics Within- and Between- Ads User Similarity Ads Click-Through Rate F-measure Ads Click Entropy 12
Symbols and Experiment Setup be the set of the n advertisements in our datasetFor each ad 2 , suppose 7 183+84+-are all the queries which have displayed or clicked 2 .Suppose the group of users who have either displayed or clicked 2 is represented by13
Symbols and Experiment Setup cont.We define a Boolean function,to show whether the user has clicked adBT aims to group users into segments of similar behaviorsk-means and CLUTO for user segmentation14
Symbols and Experiment Setup cont.The users are segmented into K segments according to their behaviorsTo represent the distribution of under a given user segmentation results where H stands for all the users in who are grouped into the kth user segmentthe kth user segment be represented by15
Evaluation Metrics Within- and Between- Ads User SimilarityThe similarity between a pair of users and is OMN the classical Cosine similarity can be utilized for the similarity computation between users16
Evaluation Metrics cont.For ad 2 , the user similarity, who clicked it, is defined as the within ads user similaritywhere [ ;< = is the number of users who clicked ad 2The between ads user similarity as17
Evaluation Metrics cont. We further define a ratio between OS2 and O asA large R score means the two ads have a large within ads similarity small between ads similaritythe more confident we are on the basic assumption of BT18
Ads Click-Through RateThe CTR of ad 2 is defined asAfter user segmentation, the CTR of 2 over user segment H iswhere RHR is the number of users in H19
F-measureA user segment H , satisfies thata segment of users who are more interested in ad 2 than other usersonly validate the precision of BT strategies in finding potentially interested usersPrecision and Recall are defined asthe classical F-measure for results evaluation20
Ads Click EntropyDefine the ads click Entropy to show the effectiveness of different BT strategiesFor ad 2 the probability of users in segment H , who will click this ad, is estimated bywe define the ads click Entropy of ad 2 as21
BT ResultsAssumption of BTshow its potential in helping online advertisingBT for Online Advertisingshow how much BT can improve ads CTRMore EvaluationMore evaluated results by the ads click Entropy and F-measure22
Assumption of BTThe average within ads user similarity over all ads The average between ads user similarity over all adsThe averaged ratio23
Assumption of BT cont.Within- and between- ad user similarity. the within ads similarity of users, which are represented by SQ, can be around 90 times larger than the corresponding between ads similarity. 99.37% pairs of ads in our dataset have the property that 24
Assumption of BT cont.To validate whether the difference between OS and O is statistically significant, we implement the paired t-test to compare the results of OS with that of .The t-test results of different BT strategies, which are all less than 0.05.25
BT for Online AdvertisingGroup all our users into 20, 40, 80 and 160 clustersCalculate `a2 over all usersCalculate its CTR over different user segments, i.e. `a2RA user segment that have the highest CTR for26
BT for Online Advertising cont.the CTR improvement degree of ad 2The average results, i.e. improvementT-test of CTR improvements by BT27
More EvaluationThe worst casesuppose we have K user segments, the users who clicked the same ad uniformly distributeAds click Entropy results28
F-measure of different BT strategies29
BT for Online Advertising cont.we present the scatter plot of the precision and recall over all the ads30
ConclusionThe users who clicked the same ad will be more similar than the users who clicked different ads; Ads CTR can be averagely improved as high as 670% over all the ads we collected if we directly adopt the most fundamental user clustering algorithms for BT; For the user representation strategies, which are defined in the definition of BT, tracking the short term user search behavior can perform better than tracking the long term user browsing behavior. 31
32~ Thank you for your listening ~

More Related Content

Behavioral targeting

  • 1. How much can Behavioral Targeting Help Online Advertising?Jun Yan, Ning Liu, Gang Wang, Wen Zhang, Yun Jiang, Zheng Chen, WWW 2009Advisor: Chia-Hui ChangStudent: Kuan-Hua HuoDate: 2010-09-071
  • 3. InstructionBehavioral Targeting (BT)BT uses information collected on an individual's web-browsing behavior, such as the pages they have visited or the searches they have made, to select which advertisements to display to that individual. Practitioners believe this helps them deliver their online advertisements to the users who are most likely to be influenced by them.3
  • 4. How much can BT help online advertising?Does BT truly have the ability to help online advertising?How much can BT help online advertising using commonly used evaluation metrics?What BT strategy works better than others for ads delivery?4
  • 5. Does BT truly have the ability to help online advertising?Users have similar search or browsing behaviorssimilar interestshigher probability to click the same adOnline users can be grouped into different user segments according to their behaviorswithin- and between- ads user similaritiesthe users who clicked the same ad can be over 90 times more similar than the users who clicked different ads.5
  • 6. How much can BT help online advertising using commonly used evaluation metrics?We propose to observe how much BT can improve ads CTR through the segmentation of users into a number of small groups for targeted ads delivery.Through studying ads CTR before and after user segmentation for ads deliveryads CTR can be improved by as much as 670% over all the ads we collected.6
  • 7. What BT strategy works better than others for ads delivery?According to the definition of Behavioral Targeting (BT), there are two strategies to represent the users behaviorWeb browsing behavior => users clicked pagesSearch behavior => search querieswe draw the conclusion that user search queries can perform several times better than user browsing behaviortracking the short term user behaviors are more effective than tracking the long term user behaviors7
  • 8. Behavioral TargetingTo represent user behavior by their page-viewsWe consider the clicked URLs of search users as their profiles All the users can be considered as a user-by-URL matrixThe classical Term Frequency Inverse Document Frequency (TFIDF) indexingeach user as a document and each URL as a term8
  • 9. Behavioral Targeting cont.Mathematically, all users are represented by a real valued matrixg is the total number of users l is the total number of URLs A user is a row of U, a real valued vector with the weight for each entry to be, 9
  • 10. Behavioral Targeting cont.We build the user behavioral profile by all terms that appear in a users queries as his previous behaviors. We can represent each user in the Bag of Words (BOW) model each term is considered as a feature.each user is represented by BOW with corresponding term frequencyWe use the same TFIDF indexing10
  • 11. Behavioral Targeting cont.Two different BT strategiesThe long term user behavior (7 days user behavior) The short term user behavior (1 days user behavior)Validate and compare four different BT Strategies1. LP: using Long term user behavior all through the seven days and representing the user behavior by Page-views; 2. LQ: using Long term user behavior all through the seven days and representing the user behavior by Query terms; 3. SP: using Short term user behavior (1 day) and representing user behavior by Page-views; 4. SQ: using Short term user behavior (1 day) and representing user behavior by Query terms.11
  • 12. Experiment ConfigurationSymbols and Experiment Setup which will be used throughout the experimentswith detailed experiment configurationsEvaluation Metrics Within- and Between- Ads User Similarity Ads Click-Through Rate F-measure Ads Click Entropy 12
  • 13. Symbols and Experiment Setup be the set of the n advertisements in our datasetFor each ad 2 , suppose 7 183+84+-are all the queries which have displayed or clicked 2 .Suppose the group of users who have either displayed or clicked 2 is represented by13
  • 14. Symbols and Experiment Setup cont.We define a Boolean function,to show whether the user has clicked adBT aims to group users into segments of similar behaviorsk-means and CLUTO for user segmentation14
  • 15. Symbols and Experiment Setup cont.The users are segmented into K segments according to their behaviorsTo represent the distribution of under a given user segmentation results where H stands for all the users in who are grouped into the kth user segmentthe kth user segment be represented by15
  • 16. Evaluation Metrics Within- and Between- Ads User SimilarityThe similarity between a pair of users and is OMN the classical Cosine similarity can be utilized for the similarity computation between users16
  • 17. Evaluation Metrics cont.For ad 2 , the user similarity, who clicked it, is defined as the within ads user similaritywhere [ ;< = is the number of users who clicked ad 2The between ads user similarity as17
  • 18. Evaluation Metrics cont. We further define a ratio between OS2 and O asA large R score means the two ads have a large within ads similarity small between ads similaritythe more confident we are on the basic assumption of BT18
  • 19. Ads Click-Through RateThe CTR of ad 2 is defined asAfter user segmentation, the CTR of 2 over user segment H iswhere RHR is the number of users in H19
  • 20. F-measureA user segment H , satisfies thata segment of users who are more interested in ad 2 than other usersonly validate the precision of BT strategies in finding potentially interested usersPrecision and Recall are defined asthe classical F-measure for results evaluation20
  • 21. Ads Click EntropyDefine the ads click Entropy to show the effectiveness of different BT strategiesFor ad 2 the probability of users in segment H , who will click this ad, is estimated bywe define the ads click Entropy of ad 2 as21
  • 22. BT ResultsAssumption of BTshow its potential in helping online advertisingBT for Online Advertisingshow how much BT can improve ads CTRMore EvaluationMore evaluated results by the ads click Entropy and F-measure22
  • 23. Assumption of BTThe average within ads user similarity over all ads The average between ads user similarity over all adsThe averaged ratio23
  • 24. Assumption of BT cont.Within- and between- ad user similarity. the within ads similarity of users, which are represented by SQ, can be around 90 times larger than the corresponding between ads similarity. 99.37% pairs of ads in our dataset have the property that 24
  • 25. Assumption of BT cont.To validate whether the difference between OS and O is statistically significant, we implement the paired t-test to compare the results of OS with that of .The t-test results of different BT strategies, which are all less than 0.05.25
  • 26. BT for Online AdvertisingGroup all our users into 20, 40, 80 and 160 clustersCalculate `a2 over all usersCalculate its CTR over different user segments, i.e. `a2RA user segment that have the highest CTR for26
  • 27. BT for Online Advertising cont.the CTR improvement degree of ad 2The average results, i.e. improvementT-test of CTR improvements by BT27
  • 28. More EvaluationThe worst casesuppose we have K user segments, the users who clicked the same ad uniformly distributeAds click Entropy results28
  • 29. F-measure of different BT strategies29
  • 30. BT for Online Advertising cont.we present the scatter plot of the precision and recall over all the ads30
  • 31. ConclusionThe users who clicked the same ad will be more similar than the users who clicked different ads; Ads CTR can be averagely improved as high as 670% over all the ads we collected if we directly adopt the most fundamental user clustering algorithms for BT; For the user representation strategies, which are defined in the definition of BT, tracking the short term user search behavior can perform better than tracking the long term user browsing behavior. 31
  • 32. 32~ Thank you for your listening ~