Influence maximization with viral product design
The disclosure includes use of a feature-aware propagation model to identify one or more features of a product and one or more person(s), or members of a social network, to target, or user, for marketing the product having the identified features. The one or more person(s) identified using the model may be the person(s), or member(s), of a social network determined to have a maximum capability, relative to other members of the social network, for influencing the members of the social network in adopting, e.g., purchasing, a product having the identified features. In addition, parameters of the model may be determined using information about the social network, user preferences, and the products and features of the products.
1. A method comprising:
selecting, by at least one computing system and using a feature-aware propagation model having a set of model parameters, a product feature set comprising a number of features of a product, the feature-aware propagation model comprising social influence, part-worth utility and adoption equilibrium predictors; and
identifying, by the at least one computing system and using the feature-aware propagation model, a number of users of a plurality of users having a higher probability, relative to other users of the plurality, to adopt the product having the product feature set, the probability that a user of the plurality adopts the product is dependent on a predicted social influence on the user, a predicted part-worth utility to the user of each feature of the product feature set of the product and a predicted adoption equilibrium determined using the feature-aware propagation model.
2. The method of claim 1 , a part-worth utility predictor for a product feature predicting an importance of the product feature to the user of the plurality of users.
3. The method of claim 1 , the probability that the user adopts the product increases as a gap between the predicted adoption equilibrium and a combination of the predicted social influence and the predicted part-worth utility increases.
4. The method of claim 1 , the probability that the user adopts the product decreases as a gap between the predicted adoption equilibrium and a combination of the predicted social influence and the predicted part-worth utility decreases.
5. The method of claim 1 , the set of model parameters of the feature-aware model comprising social influence, part-worth utility and hurdle parametric information learned from past observations of users in a social network.
6. The method of claim 1 , the set of model parameters of the feature-aware model comprising an influence strength for each of a plurality of user pairings, a part-worth utility for each of a plurality of pairings of a user of the plurality of users and a feature of the product, and a hurdle for each user of the plurality of users, the influence strength for a user pairing of the plurality of user pairings providing a measure of a strength of influence of a first user in the user pairing over the second user in the user pairing, the part-worth utility providing a measure of importance of a given product feature to a given user of the plurality of users, and a hurdle providing a product adoption equilibrium measure for the given user.
7. The method of claim 6 , further comprising:
generating, by the computing device, the set of model parameters of the feature-aware model using input comprising a social graph representing a social network, information identifying a set of products and a set of features for each product of the set of products and feature preferences and information identifying product and feature preferences of the plurality of users as members of the social network.
8. The method of claim 1 , the feature-aware propagation model determining the probability that the user of the plurality adopts the product using an assumption that the user adopts the product where a sum of the part-worth utility of each feature of the product feature set of the product exceeds the predicted adoption equilibrium determined for the user.
9. A method of claim 1 , further comprising:
identifying, by the computing device using the feature-aware propagation model having the set of model parameters, a seed set of users of the plurality of users, the identified seed set of users for influencing the number of users, identification of the seed set of users is in combination with identification, by the feature-aware propagation model, of the product feature set comprising the number of features of the product.
10. A method of claim 9 , the feature-aware propagation model determining a likelihood of the combination to maximize adoption of the product having the product's product feature set by the number of users of the plurality of users.
11. A non-transitory computer readable storage medium tangibly encoded with computer-executable instructions that when executed by a processor associated with a computing device perform a method comprising:
selecting, using a feature-aware propagation model having a set of model parameters, a product feature set comprising a number of features of a product, the feature-aware propagation model comprising social influence, part-worth utility and adoption equilibrium predictors; and
identifying, using the feature-aware propagation model, a number of users of a plurality of users having a higher probability, relative to other users of the plurality, to adopt the product having the product feature set, the probability that a user of the plurality adopts the product is dependent on a predicted social influence on the user, a predicted part-worth utility to the user of each feature of the product feature set of the product and a predicted adoption equilibrium determined using the feature-aware propagation model.
12. The non-transitory computer-readable storage medium of claim 11 , a part-worth utility predictor for a product feature predicting an importance of the product feature to the user of the plurality of users.
13. The non-transitory computer-readable storage medium of claim 11 , the probability that the user adopts the product increases as a gap between the predicted adoption equilibrium and a combination of the predicted social influence and the predicted part-worth utility increases.
14. The non-transitory computer-readable storage medium of claim 11 , the probability that the user adopts the product decreases as a gap between the predicted adoption equilibrium and a combination of the predicted social influence and the predicted part-worth utility decreases.
15. The non-transitory computer-readable storage medium of claim 11 , the set of model parameters of the feature-aware model comprising social influence, part-worth utility and hurdle parametric information learned from past observations of users in a social network.
16. The non-transitory computer-readable storage medium of claim 11 , the set of model parameters of the feature-aware model comprising an influence strength for each of a plurality of user pairings, a part-worth utility for each of a plurality of pairings of a user of the plurality of users and a feature of the product, and a hurdle for each user of the plurality of users, the influence strength for a user pairing of the plurality of user pairings providing a measure of a strength of influence of a first user in the user pairing over the second user in the user pairing, the part-worth utility providing a measure of importance of a given product feature to a given user of the plurality of users, and a hurdle providing a product adoption equilibrium measure for the given user.
17. The non-transitory computer-readable storage medium of claim 16 , further comprising:
generating the set of model parameters of the feature-aware model using input comprising a social graph representing a social network, information identifying a set of products and a set of features for each product of the set of products and feature preferences and information identifying product and feature preferences of the plurality of users as members of the social network.
18. The non-transitory computer-readable storage medium of claim 11 , the feature-aware propagation model determining the probability that the user of the plurality adopts the product using an assumption that the user adopts the product where a sum of the part-worth utility of each feature of the product feature set of the product exceeds the predicted adoption equilibrium determined for the user.
19. The non-transitory computer-readable storage medium of claim 11 , further comprising:
identifying, using the feature-aware propagation model having the set of model parameters, a seed set of users of the plurality of users, the identified seed set of users for influencing the number of users, identification of the seed set of users is in combination with identification, by the feature-aware propagation model, of the product feature set comprising the number of features of the product.
20. A computing device comprising:
a processor; and
a non-transitory storage medium for tangibly storing thereon program logic for execution by the processor, the program logic comprising:
selecting logic executed by the processor for selecting, using a feature-aware propagation model having a set of model parameters, a product feature set comprising a number of features of a product, the feature-aware propagation model comprising social influence, part-worth utility and adoption equilibrium predictors; and
identifying logic executed by the processor for identifying, using the feature-aware propagation model, a number of users of a plurality of users having a higher probability, relative to other users of the plurality, to adopt the product having the product feature set, the probability that a user of the plurality adopts the product is dependent on a predicted social influence on the user, a predicted part-worth utility to the user of each feature of the product feature set of the product and a predicted adoption equilibrium determined using the feature-aware propagation model.