710710PeopleGraphClosedGRD Proof of ConceptInnovate UK3Desk Ltd has discovered that there is no available comprehensive internet search tool for people. Some general search engines and specific membership networking and social media platforms (e.g. LinkedIn, Facebook) exist, but none work like a ‘Google for people’. The PeopleGraph Project objectives are to further investigate and prove the techniques necessary to build a viable people search engine that operates at the scale, speed and accuracy required and to further investigate the market potential. The top 300 most used platforms and sites on the web currently contain over 20bn people profiles, projected to grow to 50bn within 2 yrs. ~300 additional such platforms identified also contain significant data. The search process being developed involves data gathering (profile discovery) followed by information normalisation and decoration, resulting in identity resolution and matching. 3Desk’s techniques to process and handle the huge volumes of data required are in initial development but many critical issues of speed and accuracy remain to be solved. Key challenges in profile discovery are the quality and depth of returned data, crawler performance and optimal strategies for regular dataset refresh. Challenges in identity matching are improving the speed, accuracy and quality of outputs. Common techniques and algorithmic approaches have been largely addressed so further developments will involve innovative ground-breaking work in the following: Web crawler efficiency, Natural Language Processing (NLP), Graphing and linkage resolution on 100 billion nodes with a short, efficient processing time, Single and Relational entity matching algorithms, Data indexing, rating and ranking. Expected benefits will be the creation of new business models and services in multiple sectors, e.g. identity verification, recruitment, improved government data and services, criminal investigation, enhanced personalised marketing and market analysis, to name a few.