Big Data and Massive Analytics
0 likes173 views
This document discusses innovation for service value networks through building big data and personalizing analytics. It provides examples of networks like eBay and Facebook and innovations in areas like e-marketing, group activities, and on-demand business collaboration. The principles involve integrating personal data from various sources and developing personal services through massive analytics. The theories aim to scale connections across entire populations, individual people in networks, and through network transformations. The document also outlines an ontology for data integration and a predictive e-marketing analytics approach for customer recommendations.
Big Data and Massive Analytics
- 1. Innovation for Service value Networks Cheng Hsu Professor, RPI
- 2. Types: Internet e-Commerce Networks, Peer-to-Peer Service/Collaboration networks, Social Networks, Enterprise (Professionals) Networks, etc. Examples: e-Bay, healthcare support, Facebook, intranets… Innovations: e-marketing (customer recommendation, business chaining, etc.), group activities and special interests, on-demand business and collaboration… Technology: data integration, network analysis, clustering and statistics, personal tasks profiling… Smartness: Network-Based intelligence; i.e., population knowledge and personalization application
- 3. Principle One: Building the Big Data Integration of person-centered data along the life cycle of personal tasks and growth from all pertinent sources Principle Two: Personalizing the Big Data for services Development of personal service-oriented massive analytics to support the conduct of the personal life cycle tasks (Motto: service is the best selling) Smart Service Value Networks: possessing the ability to self-develop the Big Data and Massive Analytics for constantly evolving applications – the innovation
- 4. Theory One: Scaling the connections up to cover the entire population (business domain) – Big Data Theory Two: Scaling the connections down to serve each person (individuals of the network) – service analytics Theory Three: Scaling the connections with network transformation (hyper-networking) – business innovation All for One and One for All: a moral proposition may be an ultimate business value proposition – this is the golden rule for building Big Data and deriving Massive Analytics
- 5. 1. An ontology and metadata repository for data integration – the global information resources dictionary 2. An architecture for non-intrusive integration of massively distributed (Internet) heterogeneous data sources - the Metadatabase model 3. A core logic for predictive e-marketing analytics (e.g., the well-known customer recommendation algorithms at some e-commerce sites) A technology platform for developing the Big Data and Massive Analytics can facilitate service innovation
- 6. Enterprise Application Appluser Adminsiter User Integrates Components Defines Itemin Subject Mapped to Describes Applies Relates Meta modeled Context Contains RuleEntRel convert ActofCondof ERExist Condition Integrity Fact Action Operator Maintain Bind-Fact Calls Namedas Belongto Item Equivalent For Storedin Computes Value Software Resourses Hardware Resourses Subjectin Resides at Module of Uses Inter component Expr Loper Roper Value of Opof
- 7. The Network Metadatabase To other nodes To other nodes Proxy at Data Source Two Proxy at Data Source ThreeProxy at Data Source Four Proxy at Data Source One Mini Metadatabase
- 8. Similar Customers: 1. determine a set of defining attributes for “similarity”; 2: compute the similarity indicator, e.g., S-C(i) = ∑ w(j)a(j) for each customer i, and then group customers based on this indicator; 3: recommend the additional products that the similar customers prefer most Similar Products: use the same logic to develop a basic algorithm for using similar products S-P(i) Similar Behaviors: use the same logic to develop an algorithm from customers-products networking (compute e.g., rating/purchase indicators and regress them on attributes, by sub-groups) Adaptability can be built into the logic to make it “smart”.