Systems and methods for computer generated recommendations with improved accuracy and relevance

Abstract

The disclosed embodiments relate to a computer implemented recommender system/method which enables the computer to provide more “emotionally” relevant/connected, and therefore more likely to be successful, recommendations while minimizing dependency on historical data as well susceptibility to bias, e.g., monetary and/or sponsor based. The disclosed embodiments segment a target set of known users into groups based on similar characteristics and then associate each group with one or more unique sets of affinities derived from the social media interactions of a general population having similar affinities. Similarly, a unique set of affinity characteristics is derived, and stored in a data structure, for each of a set of recommendations. The unique sets of affinities associated with the groups are compared with the set of affinity characteristics associated with each recommendations to derive a subset thereof relevant to the users of the particular group.

What is claimed is: 
     
         1 . A computer implemented method comprising:
 defining, by a processor of a computer-based recommendation system, a set of categories, each defined based on a unique combination of values of one or more characteristics descriptive of a user;   obtaining, by the processor, for each target user of a set of target users, one or more characteristic data items, from a database coupled with the processor, each of the one or more characteristic data items comprising data indicative of a specific value descriptive of the target user;   assigning, automatically by the processor, each target user, of the set of target users, to one or more of the categories of the set of categories based on the specific values of the one or more characteristics descriptive of the target user and the combined values of the one or more characteristic which define each of the categories of the set of categories, wherein target users having similar values of the one or more characteristics are assigned to at least one of the same categories;   storing, in a memory coupled with the processor, data indicative of each of the set of categories and the target users assigned thereto;   identifying, by the processor, a representative population of users and for each user in the representative population of users, obtaining, by the processor, descriptive data, from a database of social media interactions coupled with the processor, indicative of the user's engagement with one or more of a target set of brands of goods and/or services;   compiling, by the processor for each target brand of the set of target brands, a brand affinity fingerprint comprising a set of one or more words and/or phrases present in the identified descriptive data indicative of each user's, of the population of users, engagement with the target brand;   expanding, automatically by the processor using natural language processing, the set of one or more words and/or phrases present in the identified descriptive data indicative of each user of the population's engagement with the target brand by identifying, and adding thereto, one or more synonyms of each of the set of one or more words and/or phrases, and storing the expanded compiled set of one or more words and/or phrases as the brand affinity fingerprint of the target brand in a portion of a first data structure in the memory, the brand affinity fingerprints enabling the provision of recommendations by the processor with a minimal historical dataset;   assigning, by the processor, the portion of the first data structure which stores each brand affinity fingerprint to, and storing data indicative thereof in association with, one or more of the demographic categories stored in the memory based on a relationship between one or more of the defining values of the one or more characteristics thereof and the target brand of the brand affinity fingerprint;   defining, by the processor, a set of recommendation objects, each associated with a recommendation and characterized by a recommendation affinity fingerprint comprising a set of descriptive words and/or phrases, and storing, in a portion of a second data structure in the memory, data indicative of the recommendation in association with data indicative of the recommendation affinity fingerprint;   determining, by the processor, an indicator of a degree to which the recommendation affinity fingerprint of each recommendation intersects with each of the brand affinity fingerprints by comparing the associated portions of the first data structure with the associated portions of the second data structure;   causing, by the processor for each recommendation affinity fingerprint where the indicator exceeds a threshold, the recommendation associated therewith to be provided to each of the target users assigned in the memory to the categories to which the at least one of the brand affinity fingerprints is also assigned, the causing further comprising presenting the recommendation electronically on a display coupled with the processor; and   redefining, dynamically by the processor, the set of categories and reassigning one or more target users of the set of target users to a different one or more of the categories of the set of categories based on the specific values of the one or more characteristics descriptive thereof and the result of the provided recommendation.   
     
     
         2 . The computer implemented method of  claim 1 , wherein the display is provided in a location where at least one of the target users is or will be located. 
     
     
         3 . The computer implemented method of  claim 1 , wherein the display is comprised by a mobile device. 
     
     
         4 . The computer implemented method of  claim 3 , wherein the mobile device is controlled by another person who, when prompted by the mobile device, presents the recommendation to at least one of the target users. 
     
     
         5 . The computer implemented method of  claim 1 , wherein the causing further comprises determining a geographic proximity of the target user to a location at which a product or service, which is the subject of the recommendation, is provided and providing the recommendation to the target user when the target user is proximate to the location or en route thereto. 
     
     
         6 . The computer implemented method of  claim 1 , wherein the recommendation of each of the set of recommendation objects relates to at least one of a hospitality service, a retail service or a travel service. 
     
     
         7 . The computer implemented method of  claim 1 , wherein the redefining is based on one or more of a result of the provided recommendation or additional market research. 
     
     
         8 . The computer implemented method of  claim 1 , wherein the causing is not subject to monetization or sponsor bias. 
     
     
         9 . The computer implemented method of  claim 1 , wherein the causing further comprises biasing, by the processor, one or more of the score or threshold amount to favor a recommendation related to a particular sponsor of a set of sponsors or which result in revenue for a particular vendor of a set of vendors. 
     
     
         10 . The computer implemented method of  claim 1  wherein one or more of the target users of the set of target users have self-identified to the recommendation system. 
     
     
         11 . The computer implemented method of  claim 1 , wherein the provided recommendation is tailored to each target user. 
     
     
         12 . The computer implemented method of  claim 1 , wherein the one or more characteristics comprise one or more of age, gender, marital status, home address, occupation, brand preferences, or combinations thereof. 
     
     
         13 . The computer implemented method of  claim 1 , wherein the representative population is identified based on users who have expressed interest in the service for which the recommendations are being generated. 
     
     
         14 . The computer implemented method of  claim 1 , wherein the social media interactions include likes, follows, hashtags, tags, posts, tweets, or comments. 
     
     
         15 . The computer implemented method of  claim 1 , wherein the target set of brands includes any brand engaged with by the representative population. 
     
     
         16 . The computer implemented method of  claim 1 , further comprising, prior to identifying the representative population and obtaining the engagement data, defining, by the processor, the set of target brands of goods and/or services based on relevance to the recommendations. 
     
     
         17 . The computer implemented method of  claim 1 , wherein the social media database comprises historical data, data obtained in real time, or a combination thereof. 
     
     
         18 . The computer implemented method of  claim 1 , wherein each element of the set of one or more words and/or phrases present in the identified descriptive data indicative of each user of the population's engagement with the target brand may be further associated with a weight value determined based on the particular user's influence, number of different brands engaged with, or a combination thereof. 
     
     
         19 . The computer implemented method of  claim 1 , further comprising expanding the set of descriptive words and/or phrases of each recommendation affinity fingerprint by identifying, and adding thereto, one or more synonyms of each of the set of one or more words and/or phrases, the recommendation affinity fingerprint stored in the second data structure comprising the expanded set of descriptive words and/or phrases. 
     
     
         20 . The computer implemented method of  claim 1 , further comprising processing each recommendation description to remove irrelevant content, identify linguistic units, topics or themes, and identifying, and adding thereto, one or more synonyms of each of the set of descriptive words and/or phrases of the recommendation description. 
     
     
         21 . A system comprising:
 a processor and a memory coupled therewith, the memory having stored therein computer executable program code that when executed by the processor cause the processor to:
 define a set of categories, each defined based on a unique combination of values of one or more characteristics descriptive of a user; 
 obtain for each target user of a set of target users, one or more characteristic data items, from a database coupled with the processor, each of the one or more characteristic data items comprising data indicative of a specific value descriptive of the target user; 
 assign each target user, of the set of target users, to one or more of the categories of the set of categories based on the specific values of the one or more characteristics descriptive of the target user and the combined values of the one or more characteristic which define each of the categories of the set of categories, wherein target users having similar values of the one or more characteristics are assigned to at least one of the same categories; 
 store, in a memory coupled with the processor, data indicative of each of the set of categories and the target users assigned thereto; 
 identify a representative population of users and for each user in the representative population of users, obtaining, by the processor, descriptive data, from a database of social media interactions coupled with the processor, indicative of the user's engagement with one or more of a target set of brands of goods and/or services; 
 compile, for each target brand of the set of target brands, a brand affinity fingerprint comprising a set of one or more words and/or phrases present in the identified descriptive data indicative of each user's, of the population of users, engagement with the target brand; 
 expand, using natural language processing, the set of one or more words and/or phrases present in the identified descriptive data indicative of each user of the population's engagement with the target brand by identifying, and adding thereto, one or more synonyms of each of the set of one or more words and/or phrases, and storing the expanded compiled set of one or more words and/or phrases as the brand affinity fingerprint of the target brand in a portion of a first data structure in the memory, the brand affinity fingerprints enabling the provision of recommendations by the processor with a minimal historical dataset; 
 assign the portion of the first data structure which stores each brand affinity fingerprint to, and storing data indicative thereof in association with, one or more of the demographic categories stored in the memory based on a relationship between one or more of the defining values of the one or more characteristics thereof and the target brand of the brand affinity fingerprint; 
 define a set of recommendation objects, each associated with a recommendation and characterized by a recommendation affinity fingerprint comprising a set of descriptive words and/or phrases, and storing, in a portion of a second data structure in the memory, data indicative of the recommendation in association with data indicative of the recommendation affinity fingerprint; 
 determine an indicator of a degree to which the recommendation affinity fingerprint of each recommendation intersects with each of the brand affinity fingerprints by comparing the associated portions of the first data structure with the associated portions of the second data structure; 
 cause, for each recommendation affinity fingerprint where the indicator exceeds a threshold, the recommendation associated therewith to be provided to each of the target users assigned in the memory to the categories to which the at least one of the brand affinity fingerprints is also assigned, the causing further comprising presenting the recommendation electronically on a display coupled with the processor; and 
 redefine, dynamically, the set of categories and reassigning one or more target users of the set of target users to a different one or more of the categories of the set of categories based on the specific values of the one or more characteristics descriptive thereof and the result of the provided recommendation. 
   
     
     
         22 . The system of  claim 21 , wherein the display is provided in a location where at least one of the target users is or will be located. 
     
     
         23 . The system of  claim 21 , wherein the display is comprised by a mobile device. 
     
     
         24 . The system of  claim 23 , wherein the mobile device is controlled by another person who, when prompted by the mobile device, presents the recommendation to at least one of the target users. 
     
     
         25 . The system of  claim 21 , wherein the computer executable program code is further executable by the processor to cause the processor to determine a geographic proximity of the target user to a location at which a product or service, which is the subject of the recommendation, is provided and providing the recommendation to the target user when the target user is proximate to the location or en route thereto. 
     
     
         26 . The system of  claim 21 , wherein the recommendation of each of the set of recommendation objects relates to at least one of a hospitality service, a retail service or a travel service. 
     
     
         27 . The system of  claim 21 , wherein the redefinition is based on one or more of a result of the provided recommendation or additional market research. 
     
     
         28 . The system of  claim 21 , wherein the provision of the recommendation is not subject to monetization or sponsor bias. 
     
     
         29 . The system of  claim 21 , wherein the computer executable program code is further executable by the processor to cause the processor to bias one or more of the score or threshold amount to favor a recommendation related to a particular sponsor of a set of sponsors or which result in revenue for a particular vendor of a set of vendors. 
     
     
         30 . The system of  claim 21 , wherein one or more of the target users of the set of target users have self-identified to the recommendation system. 
     
     
         31 . The system of  claim 21 , wherein the provided recommendation is tailored to each target user. 
     
     
         32 . The system of  claim 21 , wherein the one or more characteristics comprise one or more of age, gender, marital status, home address, occupation, brand preferences, or combinations thereof. 
     
     
         33 . The system of  claim 21 , wherein the representative population is identified based on users who have expressed interest in the service for which the recommendations are being generated. 
     
     
         34 . The system of  claim 21 , wherein the social media interactions include likes, follows, hashtags, tags, posts, tweets, or comments. 
     
     
         35 . The system of  claim 21 , wherein the target set of brands includes any brand engaged with by the representative population. 
     
     
         36 . The system of  claim 21 , wherein the computer executable program code is further executable by the processor to cause the processor to, prior to the identification of the representative population and obtaining the engagement data, define the set of target brands of goods and/or services based on relevance to the recommendations. 
     
     
         37 . The system of  claim 21 , wherein the social media database comprises historical data, data obtained in real time, or a combination thereof. 
     
     
         38 . The system of  claim 21 , wherein each element of the set of one or more words and/or phrases present in the identified descriptive data indicative of each user of the population's engagement with the target brand may be further associated with a weight value determined based on the particular user's influence, number of different brands engaged with, or a combination thereof. 
     
     
         39 . The system of  claim 21 , wherein the computer executable program code is further executable by the processor to cause the processor to expand the set of descriptive words and/or phrases of each recommendation affinity fingerprint by identification, and addition thereto, of one or more synonyms of each of the set of one or more words and/or phrases, the recommendation affinity fingerprint stored in the second data structure comprising the expanded set of descriptive words and/or phrases. 
     
     
         40 . The system of  claim 21 , wherein the computer executable program code is further executable by the processor to cause the processor to process each recommendation description to remove irrelevant content, identify linguistic units, topics or themes, and identifying, and adding thereto, one or more synonyms of each of the set of descriptive words and/or phrases of the recommendation description. 
     
     
         41 . A system comprising:
 means for defining a set of categories, each defined based on a unique combination of values of one or more characteristics descriptive of a user;   means for obtaining for each target user of a set of target users, one or more characteristic data items, from a database coupled with the processor, each of the one or more characteristic data items comprising data indicative of a specific value descriptive of the target user;   means for assigning, automatically, each target user, of the set of target users, to one or more of the categories of the set of categories based on the specific values of the one or more characteristics descriptive of the target user and the combined values of the one or more characteristic which define each of the categories of the set of categories, wherein target users having similar values of the one or more characteristics are assigned to at least one of the same categories;   means for storing data indicative of each of the set of categories and the target users assigned thereto;   means for identifying a representative population of users and for each user in the representative population of users, obtaining, by the processor, descriptive data, from a database of social media interactions coupled with the processor, indicative of the user's engagement with one or more of a target set of brands of goods and/or services;   means for compiling, for each target brand of the set of target brands, a brand affinity fingerprint comprising a set of one or more words and/or phrases present in the identified descriptive data indicative of each user's, of the population of users, engagement with the target brand;   means for using natural language processing to expand, automatically, the set of one or more words and/or phrases present in the identified descriptive data indicative of each user of the population's engagement with the target brand by identifying, and adding thereto, one or more synonyms of each of the set of one or more words and/or phrases, and storing the expanded compiled set of one or more words and/or phrases as the brand affinity fingerprint of the target brand in a portion of a first data structure in the memory, the brand affinity fingerprints enabling the provision of recommendations by the processor with a minimal historical dataset;   means for assigning the portion of the first data structure which stores each brand affinity fingerprint to, and storing data indicative thereof in association with, one or more of the demographic categories stored in the memory based on a relationship between one or more of the defining values of the one or more characteristics thereof and the target brand of the brand affinity fingerprint;   means for defining a set of recommendation objects, each associated with a recommendation and characterized by a recommendation affinity fingerprint comprising a set of descriptive words and/or phrases, and storing, in a portion of a second data structure in the memory, data indicative of the recommendation in association with data indicative of the recommendation affinity fingerprint;   means for determining an indicator of a degree to which the recommendation affinity fingerprint of each recommendation intersects with each of the brand affinity fingerprints by comparing the associated portions of the first data structure with the associated portions of the second data structure;   means for causing, for each recommendation affinity fingerprint where the indicator exceeds a threshold, the recommendation associated therewith to be provided to each of the target users assigned in the memory to the categories to which the at least one of the brand affinity fingerprints is also assigned, the causing further comprising presenting the recommendation electronically on a display coupled with the processor; and   means for redefining, dynamically, the set of categories and reassigning one or more target users of the set of target users to a different one or more of the categories of the set of categories based on the specific values of the one or more characteristics descriptive thereof and the result of the provided recommendation.