Adaptive collaborative matching
AssigneeRealm IP, LLC
InventorsJULIE L. FAUPEL, HUNTER ALBRIGHT, EDWARD DOMBROWER
Life of this patent
Estimating the term…
Abstract
An adaptive collaborative platform applies various machine learning techniques to correlate potential purchasers with high-value articles of property that may be of interest. Attributes, characteristics, preferences, and the like of a potential purchaser are scored against attributes and features of articles. The platform learns from interaction by the agents and the potential purchasers and adapts to become more attuned to the desires and lifestyle of purchasers and to gain more and more pertinent information from the listing agents regarding high-value articles, so as to ultimately to arrive at a better match between a high value article for sale and a likely purchaser.
Claims17
A computer-implemented method for adaptive collaborative matching among a plurality of heterogeneous entities, the method comprising:
- collecting, by at least one processor, structured data and unstructured data associated with a plurality of entities, the plurality of entities including at least first entities of a first type and second entities of a second type;
- normalizing, by a normalization engine executed by the at least one processor, the structured data and the unstructured data into a common structured schema;
- deriving, by a tag derivation engine, a plurality of factors from the normalized data, each factor representing a measurable characteristic of an entity;
- grouping selected factors into a plurality of tags, each tag comprising a weighted combination of two or more factors;
- assigning, for each entity,
- (i) a tag score for each tag, and
- (ii) a tag confidence value indicative of reliability of the tag score;
- computing, by a lifestyle engine, a plurality of behavioral profile scores for each entity based on weighted combinations of tag scores and tag confidence values;
- bi-directionally correlating, by a matching engine, behavioral profile scores of entities of the first type with behavioral profile scores of entities of the second type to generate a ranked set of candidate correlations;
- presenting, via a user interface, at least a subset of the ranked candidate correlations;
- receiving feedback data associated with at least one presented correlation;
- and
- modifying at least one of
- (i) factor weight assignments,
- (ii) tag compositions,
- (iii) tag confidence calculations, or
- (iv) behavioral profile scoring rules
- based on the received feedback data to produce a refined matching mode.
An adaptive collaborative matching system comprising:
- at least one processor;
- a non-transitory computer-readable storage medium storing instructions that, when executed by the at least one processor, cause the system to:
- normalize structured and unstructured data associated with a plurality of entities;
- derive weighted factors from the normalized data;
- form tags from combinations of the weighted factors;
- assign tag scores and tag confidence values to each entity;
- generate behavioral profile scores for each entity based on weighted combinations of tag scores and tag confidence values;
- correlate behavioral profile scores of entities of a first class with behavioral profile scores of entities of a second class to generate ranked correlation results;
- generate a first matching model;
- receive user-generated feedback associated with correlation results;
- and
- generate a second matching model that modifies at least one of factor weighting, tag composition, profile scoring, or correlation ranking relative to the first matching model.
A non-transitory computer-readable storage medium storing instructions that, when executed by at least one processor, cause the processor to:
- construct a structured data representation for each of a plurality of entities from heterogeneous data sources;
- derive lifestyle-representative tags from weighted factors associated with the structured data representation;
- generate multi-dimensional behavioral profile vectors for each entity;
- compute similarity metrics between behavioral profile vectors of dissimilar entity types;
- produce ranked correlation outputs based on the similarity metrics;
- receive transaction-based or interaction-based outcome signals;
- and
- iteratively refine at least one parameter of the similarity metrics or tag derivation process based on the outcome signals.
Dependencies are inferred from the claim text.