(text/JJT)

is based on the analysis of the large data is convenient, people do more and more accurate in information matching. Consumers are lack of goods, missing is the right one. Nerd founder Han Sha internship at university found an interesting phenomenon: on the one hand, the demand for their consumers don’t know much, on the one hand, and guide to stay at a respectful distance from sb. Selection dilemma means consultant market, traditional way recommended quite clear, easy to make the person produces resistance. Based on the above reasons, Han Sha produced the idea of personalized recommendation service. In China, the book is cheap, classification, easy to expand, more is less affected by the goods. Therefore, founding team choose books recommended as the breakthrough point.

open a Nerd, a drop-down menu, the user needs to answer some similar “are you after a few zero”, “want to buy which brand of mobile phone”, “average working hours” problem. Through these questions, Nerd for users to establish a preliminary portrait. Then using a mobile phone number registered user can select or use the existing SNS accounts (weibo, micro letter, watercress). The drop-down menu, Nerd can alternate an article and recommend a book. Currently recommended books are based on data from douban and jingdong, recommend articles to grab on the Internet. Users can “like”, “dislike” button to recommend the result feedback.

specific to the algorithm level, Nerd hierarchical matching algorithm completely independent development, implemented by three steps, in four aspects.

can be summarized into three steps: first, through the analysis of user’s character and key information semantic analysis, the second, based on keyword matching books, articles, video, three, will push the matching result to the user.

four aspects respectively is: the user information acquisition, content, characteristic, content, characteristic information collection, the user preference.

user information acquisition with active and passive two modes. Including direct questions, active users offline behavior acquisition; Passive including SNS information collection, APP local behavior sampling and other platforms.

the content information collection is divided into two aspects of books and information. Books, including “category, involves the contents, characters, involves the books and goods. Information books including books content abstract, evaluation, contents in this paper. After completing the two acquisition according to the calculated Nerd technology, output user preferences features, content, these two items.

user preference characteristics including brand preference, preference, education background, social crowd behavior preference. Features include style features, associated characteristics, way. The content of the processed data will form the user, such as: powerful, imagination, knowledge, open, such as planning, traditional keywords. These keywords will pass character content matching function to calculate, the matching information through ranking module screening, screening results were finally pushed to the user’s hands.

compared to today’s headlines, douban reading and other similar competing goods, Nerd has the following advantages.

1, pure. Not been kidnapped by the circle of friends, away from the hustle and bustle of SNS. Nerd in douban as the Kindle to the device. Focus will be more profound.

2, pure and fresh and have depth. although pure science founding team believe that “art is a kind of disease,” but good UI design won the favour of literary youth. Compared to today’s headlines in order to obtain the information for the purpose of reading, Nerd lay particular stress on the depth of the user personalized reading more.

3, accurate. Nerd will record the user operation behavior in the application, use the longer, push the result more accurate.

team, graduated from Beijing forestry university founder Han Sha information management and information system specialty, a former flash silver senior product manager, responsible for the development of 91 ious. Co-founder and CTO yu ze, the master of physics of the university of Massachusetts, deputy chief, flashing silver before data model for flash silver data model. More than software development engineers, was the state administration of taxation information platform, head of research and development. Front-end developer Feng Di, Seth was the east software development engineers, dongfeng DFPC, head of global view and the data collection and analysis platform. Product design huber jiyang, graduated from Beijing university of science and technology industrial design professional, responsible for the large data and connection machine.

Nerd online, at the end of June this year in less than a month of time has more than 10000 registered users. It is understood that Nerd recently has received one million seed round of investment, investors for individual investors.