System and method for Online match-making recommendation
Publication Date: 2016-Nov-02
The IP.com Prior Art Database
A novel iterative matching system blending platform features, online sources and historical interactions.
Page 01 of 2
System and method for Online match -making recommendation
Disclosed is a system that assists in online matching recommendation of people by blending platform features, online sources and historical interactions to generate reliable and complete user profile based on the combination of the following steps:
1. An end-to-end intelligent system that suggests "Overall User Profile Strength" and also Confidence and Importance score for each of the User Attributes based on the specific source of information used to gather that profile and continues learning model to learn reliability of a given source for an attribute
•For Example: Based on the continuous learning model the system first learns which attributes of a particular user are critical to judge his personality. System also learns a subset of attributes to be gathered directly from user by asking selective questions. It relies on automated methods for offline mining of his personality for remaining set of identified attributes.
2. Periodic update of the User Attributes List based on analysis of the historical interaction data of the matching recommendations and the final outcome
•For example: The system uses the history of past interactions (of the current user or the general pool of users) to understand which of the interactions had failed and further understand the cause of the failure and map them back to behavioral traits of the participating individuals by
•Analyzing the history of interaction (e.g. chatting history etc) to understand which kind of interaction had provoked a disagreement in the past •Based on the analysis, tries to learn attributes that can be captured as part of user profile early-on
Differences from Prior Work :
Our system is different from previous works in the field because of the following reasons
Our system learns important attributes and continuously updates them in a systematic way (learnt from past history) and not purely predetermined.
Our system determines the best source to validate the attribute values and not relies explicitly on user entries.
Our system evaluates user profile based on entry values and their confidence.
Main Idea :
The solution is a two step process
Identify attributes and sources to compute user social profile strength.
Periodic update of...