System and method for creating a preference profile from shared images
A method includes obtaining from an online social media site a plurality of instances of images of objects associated with a person; analyzing with a data processor the plurality of instances of the images with a plurality of predetermined style classifiers to obtain a score for each image for each style classifier; and determining with the data processor, based on the obtained scores, a likely preference of the person for a particular style of object. The plurality of instances of images of objects associated with the person can be images that were posted, shared or pinned by person, and images that the person expressed a preference for. In a non-limiting embodiment the object is clothing, and the style can include a fashion style or fashion genre including color preferences. A system and a computer program product to perform the method are also disclosed.
1. A method, comprising:
obtaining from an online social media site a plurality of instances of images of objects associated with a person;
analyzing with a data processor the plurality of instances of the images with a plurality of predetermined style classifiers to obtain a score for each image for each style classifier;
determining with the data processor, based on the obtained scores, a likely preference of the person for a particular style of object; and
based on a detected evolution over time of the person's likely preference for a particular style of the object, predicting the person's next new preference for a style of the object so as to at least make a recommendation to the person in accordance with the predicted next new style preference of the object.
2. The method as in claim 1 , where the likely preference of the user is used to at least recommend to the person another person having a similar preference.
3. The method as in claim 1 , where the likely preference of the user is used to segment a plurality of persons into segments having similar preferences.
4. The method as in claim 1 , where analyzing comprises:
obtaining a classification score vector for each image for each classifier;
aggregating the classification score vectors over all of the images; and
deriving the likely preference of the person for a particular style of the object as a probability distribution over a plurality of pre-defined object styles.
5. The method as in claim 1 , comprising an initial step of generating the plurality of predetermined classifiers by performing feature extraction on a plurality of positive style examples and negative style examples of annotated images, and performing machine learning on the extracted features to obtain a classifier specific for each type of style.
6. The method as in claim 1 , where the plurality of instances of images of objects associated with the person are images that were posted, shared or pinned by person, and images that the person expressed a preference for.