How to speed up the efficiency of intelligent ente

2022-08-25
  • Detail

Zhujian Intelligence: how to speed up enterprise efficiency? The deep capabilities of the knowledge map support the re upgrading of intelligent applications

the previous generation of enterprise service robots of a leading enterprise in the logistics industry have poor semantic understanding ability, low interception rate and no central control platform, which not only failed to achieve the ideal effect of intelligent services, but also cost a lot of human operation and maintenance. In order to improve this situation, the logistics giant cooperated with Zhujian intelligence. In a short period of 10 days, the project team of Zhujian embodied its super AI combat ability. With the one-stop emotional interaction artificial intelligence platform of BOT factory, combined with Gemini (knowledge factory) knowledge engineering platform, it automatically built a general and industry knowledge map, quickly accumulated the knowledge base of the logistics industry, combined with natural language understanding and multiple rounds of dialogue, Build an enterprise question and answer service robot that can provide accurate and intelligent services

the launch of the robot coincided with the double 11. During this period, the service machine of this leading logistics enterprise currently has an average daily conversation volume of 700000, with the highest daily conversation volume of 830000. It takes an average of 2 minutes for a customer to be served by manual seats, and it takes at least 3500 manual seats to ensure the quality of service

the enterprise Q & a service robot built by Zhujian helped him solve 95% of the problems alone, and the semantic understanding ability and interception rate were significantly improved, and the number of turns to manual was significantly reduced. The service information covers more than 10 business areas, such as user services, platforms, cooperative franchising, market activities, operation guarantee, value-added services, and supports multi-channel access such as app, Alipay, and applet, adding new impetus to the speed miracle of the express industry

01 quality! Knowledge map with full coverage of industry knowledge

in the Intelligent Interaction scenario of applying dialogue robots, enterprises often pay attention to two questions for the sake of cost reduction and efficiency increase: can robots replace human beings to answer questions? Is the robot's question answering good enough and intelligent enough? "Whether to answer questions" is determined by the range of robot knowledge, and "whether the question is intelligent enough" is determined by the quality of knowledge mastered by the robot

is the upper application smart enough? First look at the quality of knowledge base and knowledge map. Without the coefficient error on the dashboard, the value of an enterprise depends on its knowledge assets. As an important technology of cognitive intelligence, the quality of knowledge map determines the effect of knowledge application

take the logistics scenario for example. Users of logistics enterprises often ask questions about consignments, such as: can I send watermelons? If you encounter a robot that can't understand human words, you can only understand that the user is consulting about sending mail. You often push a general rule card and rush things, which doesn't directly solve the user's consultation about the specific item watermelon. Users may not be satisfied with this answer, and then request to transfer to human services or exit the dialog directly

the robot can't provide satisfactory consulting services because of the lack of common sense and knowledge about watermelon in its knowledge system. In the traditional question and answer system of semantic matching architecture, a large number of similar questions need to be configured, and the model can give correct answers after repeated training. This method of relying on large-scale data annotation not only consumes manpower, but also has scattered management and can't be unified

and bamiya Gemini's intelligent knowledge base already has prior knowledge related to the industry. Based on this, an intelligent question answering robot can accurately identify the entity watermelon. Through its subordinate relationship, watermelon and fruit are fresh, which can be identified as fresh products, and then match the consignment rules and policies of specific fresh products in logistics industry knowledge, combined with multiple rounds and task logic, It can directly generate an accurate rule reply for watermelon and further deal with subsequent service requirements

Zhujian Gemini knowledge engineering platform covers the knowledge map of versatility and industry knowledge. The semantic network formed by this support can deal with more complex contexts, make the upper application service robot smoother and more intelligent, and realize fine and deep semantic understanding and process automation

02 speed! Automatic AI platform product capability

the key to ensuring the rapid implementation of AI projects is the automatic AI platform products with low code and rapid secondary development developed by Takeshi

Zhujian Gemini knowledge engineering platform can unify all kinds of documents, FAQs and business standard processes, whether structured, semi-structured or unstructured data, for knowledge modeling, extraction, fusion and storage, automatically build a visual knowledge map, clearly show the relationship between data and knowledge, and integrate knowledge acquisition, editing, management and search, Form an industry knowledge base that traders and investors are trying to reserve a large number of non-ferrous metals to support intelligent applications before the further depreciation of the RMB

in addition, the BOT factory one-stop enterprise level emotional interaction artificial intelligence platform in Zhujian can quickly build business scenario robots through preset model algorithms and pre training intentions. For example, in the cooperation with the logistics giant, the robot dialogue and data analysis were carried out through the platform to locate the problem of high-frequency labor transfer of wear-resistant peek, a special engineering plastic robot with good comprehensive performance in the past, and the solutions were put forward. Through combing the customer business, multiple business multi round scenarios such as timeliness query, forwarding and return, claim settlement scenario, express dispatch and so on were established, and the proportion of labor transfer decreased significantly after going online

03 efficiency! Enterprise level knowledge base, unified management of knowledge

Zhujian Gemini knowledge base is an enterprise level knowledge management platform, which shares and manages knowledge in a unified platform. New knowledge and user portrait knowledge generated in project documents and business scenarios are constantly updated and automatically collected in Gemini intelligent knowledge base, avoiding the hidden dangers of decentralized management of knowledge by various departments, It can also provide external customer service to robots receiving various business scenarios, and also provide knowledge search and other applications for internal employees of the enterprise, so as to shorten the time of knowledge acquisition and improve efficiency

the unified knowledge base, combined with the central control platform, has the function of robot incoming flow distribution, which can realize the on-line task scene by channel and in proportion to gray scale, and effectively control risks; You can also compare the service effects of robots in different task scenarios and process versions of different task scenarios online, and find the service differences by analyzing the real dialog data, so as to improve the user experience

Copyright © 2011 JIN SHI