In big data analytics, Huawei focuses on three key sectors: telecommunications, enterprises and consumers. With FusionInsight, Huawei has built a hugely powerful collection of big data solutions that can create greater value for our customers and significantly enhance the efficiency of work and life.
Big data is a hot topic. In fact, Huawei’s founder Ren Zhengfei realized the importance it would have in the future. A number of years ago, he indicated that the future of ICT would be decided by big data, and proposed the Noah’s Ark Lab for innovations in machine learning, data mining, and artificial intelligence. Ren did not use the term "big data" at the time of course; the expression he used was "information flood". He once discussed with us how Huawei could meet the challenges of this information flood, as well as the future of these technologies. Today, these are the challenges and topics that the whole industry is tackling. This article is about Huawei's achievements in big data analytics.
FusionInsight: Integrating Huawei's leading achievements in big data
In 2014, on the back of many years of research and development in big data analytics, Huawei launched its big data product– FusionInsight, a platform that contains a range of big data solutions for different scenarios. Huawei's R&D teams and product teams jointly developed the core technologies and algorithms to support the product. With the launch of FusionInsight, Huawei will be able to provide better products and services for its customers.
During FusionInsight's development, two key terms continually cropped up, reflecting Huawei's strategy and capabilities in big data analytics. They were "application-driven" and "algorithm-centric". As a leading ICT solutions provider, Huawei has an excellent understanding of the needs of different types of customers and the characteristics of applications from various sectors. Huawei also possesses a very strong capability at developing algorithms for big data analytics. In addition, it has many outstanding ICT platforms. These are all Huawei's strengths in the field of big data.
Huawei's big data analytics focuses on three key sectors: telecommunications, enterprises and consumers. After analyzing the key applications in these three areas, Huawei divides the data from all these applications into six types: tabular, stream, graph, text, temporaspatial, and multimedia. Targeting the characteristics of these key applications and data types, Huawei has developed a large number of advanced algorithms and tools in numerous important areas, including deep learning, stream data mining, graph data mining, and large-scale machine learning. Huawei has fully leveraged its existing platforms and solutions to bring these together to build the exceptionally powerful platform and solutions for big data analytics – FusionInsight.
Structurally, FusionInsight has three layers: a system layer at the bottom, a tool/algorithm layer in the middle, and at the top an application layer which contains a variety of applications.
Different types of data analytics require different tools, and FusionInsight offers a large number of powerful tools and algorithms at the tool layer, consisting of two main types – machine learning and data mining. These tools include a search engine, a recommendation engine, a graph data mining toolkit, and a stream data mining toolkit.
At the application layer, Huawei focuses on carriers, enterprise and consumer applications. In the carrier sector, Huawei offers five key applications of particular importance for telcos: intelligent customer relationship management (CRM), big data services, future network architecture, intelligent network maintenance, and intelligent network planning and optimization.
In the consumer sector, the key applications are the intelligent applications store and intelligent information management; while in the enterprise sector, the key applications are intelligent banking and intelligent management.
Success in industry applications
2014 was a very fruitful year for Huawei in the field of big data analytics. It marked the official release of FusionInsight, a product of many years of collaborative effort on the part of numerous departments in the company. In addition, a range of powerful big data analytics technology and products were successfully implemented for different customers and in various scenarios. The following examples demonstrate how FusionInsight has been utilized to help customers solve problems in their businesses.
The first example pertains to applications for the telecom industry. Operators have accumulated a huge amount of data during their operation, including the so-called OSS and BSS data. In the past, it was very difficult for operators to make churn predictions using big data. With FusionInsight, Huawei helps operators build big data platforms that leverage Huawei's technical expertise to produce effective customer loss forecasts that help increase customer retention from 15% to 35%. The main technology used is automatic user modeling, which effectively captures key customer attributes. These are then employed to forecast how likely a user will be to terminate service contracts, thus helping operators retain customers.
The second example relates to applications for the financial sector. Commercial banks usually rely on transaction data to build customer models to provide personalized services. Such models are critical for boosting the effectiveness of promotional activities. Huawei FusionInsight's intelligent banking applications leverage big data to help banks create more accurate user models with thousands, even millions, of features. With this solution, Huawei helped a commercial bank in China increase the success rate of their target promotion sixfold.
In the consumer market, FusionInsight also offers an extensive range of in-depth applications. For example, all Huawei phones released since September 2014 offer the intelligent help feature. With this feature, users can ask questions through voice input or key touch input about how to use Huawei smartphones. The accuracy of the answers provided is as high as 90%.
This new function is powered by Huawei's big data analytics technology. The biggest challenge for the intelligent help function is that the answers provided by the phone do not match users' questions. To address this, Huawei is continuing to make advances in machine matching technology to provide matches for answers based on semantics (sentence meaning), and to increase the accuracy of answers and help users better solve problems. More Huawei smartphone users are now using this function, with over 100,000 questions collected on the cloud every day.FusionInsight is a very large product portfolio, as is demonstrated from these examples. It includes many technologies that will only continue to improve and feature more new functions, creating greater value for customers and enhancing their efficiency in work and daily life.