Three minutes to understand the application of intent search technology in security field

Intention search originated from the Internet industry search engine tools. With the increasing amount of Internet information, it is more and more difficult to find information quickly and accurately. The main reason is that search engines cannot understand the user's true query intent, so machine learning is inherited. Application research in this area, such as algorithms and BP neural network methods, has quickly attracted attention. Lam Wai uses case-based machine learning and query feedback techniques to automate the classification of text and use it for text retrieval. Experiments have proven its superiority. Mandala Rila uses weighting methods to implement query expansion from multiple types of information, and features extraction based on matrix algebra-based topic extraction algorithms, combined vocabulary and index texts explored by ChakrabarTI are all efforts in this regard.

But the most promising is the search system developed by Doug Lenat, which builds a common sense library and an inference engine to talk to users, thus realizing the understanding of the user's true intentions. Feigenbaum, the father of knowledge engineering, praised him for creating the "Age of SemanTIcs". But the premise is to build a huge library of common knowledge encyclopedias, which is very difficult, and does not analyze the user's potential intentions. One way to construct a large knowledge base is to construct it automatically from the network. Craven and Lesser proposed the system principle of information acquisition, the structure and implementation method of the knowledge base. Choi implements a specific automated construction system that discovers text from the web and falls into the local database for query, which is equivalent to a smart proxy. Alsaffa did a similar exploration, using an expert system to achieve automatic conversion between user-like expressions and system-required expressions. Intent search is an automated search that is intelligent based on these technologies. Although the intent search originated in the Internet industry, it has been widely used in other industries, such as the security industry and the Internet of Things technology.

The Internet industry was first started by search engines, and the current data is also through the search engine to achieve data exchange and delivery process. The reason why users generate search behavior is often to solve the task when they encounter unfamiliar concepts or problems, which leads to the demand for specific information. After that, the user will gradually form the query words of the second-speed demand in the mind, and the query will be Submitted to the search engine, and then browse the search results. If the search results can not completely solve the user's information needs, the query will be rewritten according to the inspiration of the search results, in order to more accurately describe their information needs, and then reconstruct the new Query the requirements and submit the search engine, thus forming a closed loop of user and search engine interaction until the search results have solved their needs or tried several times without results. It can be seen from the above process that there is a great uncertainty between the user's generation of information requirements and the final formation of user queries. Users may not be able to find suitable query words from the beginning. Even if they are found, there may be query words. A situation in which the information needs are fully described, that is, there is a problem of information loss in the process of forming a query. Therefore, the query rewriting in the subsequent loop is a process in which the user gradually clarifies the search requirements.

Each search request sent by the user implies a potential search intent. If the search engine can automatically find out the user's search intent based on the query vocabulary, and then provide different search methods for different intents, it will be more in line with the user's intention. The search results are at the forefront, which will undoubtedly increase the search experience of search engine users. At present, search engines have partially implemented this search mode. For example, when users search for "Beijing weather", they will actively list the temperature of the day and other conditions at the top of the search results.

Intention search is based on the current unorganized, heterogeneous, distributed and dynamic characteristics of information on the internet and the shortcomings of existing search. It is used to solve the "information overload" and "resources" faced by existing information retrieval systems. And the actual problems such as the personalized requirements of search results, to achieve personalized personal information services. On the basis of the concept of strengthening the "intelligent" direction, the search engine launches intelligent navigation, concept search and personalized search functions based on automatic classification and automatic clustering, so that the search engine can fully understand the user's intention to search. This is the intention. Search for a landing application model for search engines.

At present, most search engines improve the accuracy and hit rate of user search requests through user registration and user personalized configuration (using a cookie mechanism, creating profiles for users, etc.) to achieve a certain degree of personalized search service. This method requires the user to register personal information on the server, but this may result in the disclosure of certain private information of the user. The currently accepted search engine intent search method is to adopt the feature acquisition method. Feature acquisition methods are divided into two categories. The first type of method can be called a prior method. This method uses the characteristics of the query itself to represent the query before the query is submitted to the search engine, such as characteristic words, words and words that represent specific requirements. The relationship between words, the part of speech and the choice of words, the statistical information in the corpus, etc.; the second type of method can be called the after-the-fact method, which uses the relevant data submitted by the query to the search engine to obtain the query. Characteristics.

The intent search is divided into two parts: the intent analysis and the analysis extension. Intention analysis is an effective way to solve the problem of “information overload” and “resource fascination”. By collecting, processing and sorting out the information resources required by users, it is possible to sort out and order the professional information resources on the network to provide users with information. Intelligent information service. The specific steps of the intent analysis and analysis extension are as follows:

(1) Intention analysis. 1 According to the user's answer to the question, retrieve the relevant information in the knowledge base to provide a direct answer. 2 The original query of the input is organized and arranged according to its concept content to extract relevant conceptual information and category knowledge. Then, the query vector is modified by intental reasoning such as semantic association. 3 The query vector is submitted to the user for confirmation and editing. 4 According to the user's editing result, the intent can be inferred again until the user is satisfied. 5 Record the results of this intent reasoning into the knowledge base for future intent reasoning.

(2) Analysis extension. Based on the results of the intent analysis, the original query vector is updated to form a new query vector. The method is to combine the application domain knowledge with the index, relevance, estimation and query expression to implement the query expansion, that is, the query index also includes the part of the query word that does not appear in the user query. The selection strategies for query terms required for query expansion are: 1 non-independent classes. Non-independent words refer to words that have a greater relevance to the query term. Query expansion finds related words, phrases, sentences, paragraphs/texts from the knowledge base through knowledge base reasoning. 2 interactive selection. The user decides the last query word from the candidate words obtained through the above strategy.

The shocking global "Malaysia Airlines" incident in 2014 has touched the hearts of countless Chinese people. The Pan Pacific countries have provided a lot of help to search for lost flights, but because of the vast variety of search methods, the information reaches a massive level. A variety of problems, such as the discontinuity of information sources, have led to delays in search results. By carefully analyzing the search methods, it can be found that this type of search is similar to the search engine intent search in the Internet industry. The similarity is that data collection, screening, filtering, classification, and integration are not a single action. But a whole set of coherent processes.

The Internet of Things industry itself has characteristics such as diversity of data forms and massive amount of information due to the variety of sensors. If the large-text environment still uses the past full-text search method for single search, it will not be able to meet the intelligent needs of users in the future. In response to this challenge, Academician Wu Manqing of the Chinese Academy of Engineering at the 2014 China Big Data Summit published a speech entitled “There are traces, there is information, and the intentional search of big data in the anti-terrorism environment.” Wu Academician It describes the characteristics of the application of anti-terrorism action in the new era: legal guarantee, human-centered, high-tech means, international cooperation, and the use of virtual space information society radar to cite intent search to counter-terrorism activities and enhance anti-terrorism capabilities.

Based on the basic concept of “there are traces, the information is connected”, the model of behavioral events and thought events in the cyberspace is constructed, and the traces and contact information of the human beings are collected in the physical space and the virtual space, focusing on the characters or events. The chain of contacts, in order to achieve the various intents in the cyberspace, complete a comprehensive description of the origin of the intent, the path of communication and the trend of development, to achieve a comprehensive search and analysis of the thoughts and actions of individuals and organizations. The Internet of Things industry has a large amount of video, audio and image data. The corresponding intelligent algorithms can extract structured information for classification. The combination of these structured information and intent search technology can help users quickly predict things and events. Probabilities and trends.

Due to the grim domestic anti-terrorism situation, big data search needs to actively respond to the threat of terrorism in the new era, based on the existence and intention of terrorism, using big data technology to conduct in-depth correlation analysis, to achieve accurate identification of terrorists and early warning of terrorist incidents. Strongly enhance the core capabilities of counter-terrorism and stability. As the leader of the security industry, Hikvision has already begun to address the concept of the Internet of Things. It is believed that in the near future, the intent search application may be the first to appear at Hikvision.

This paper first explains some of the sources and basic concepts of intent search, and then discusses some specific development processes, application models and technologies of the Internet industry that generate the concept of intent search. Finally, the specific combination of security industry and intention search technology is combined. The form has been discussed. Through the predictive analysis of the Malaysian Airlines loss incident, the reader is told how to search for how to play its role in security, anti-terrorism and other fields, and how to closely integrate with the Internet of Things.

2.5MM Wire To Board Connectors

2.5MM Wire To Board Connectors.

Connectors can be classified into three basic types: line to line, line to board and board to board. These three types of connections are not entirely different. There are two reasons for this type of overlap. First of all, the design of the same connector only needs to be redefined after slightly changing the connection mode, that is to say, it becomes a new design scheme that can be applied to another type of connection mode; secondly, a cable can be installed with a line to line connector at one end and a line to board connector at the other end during assembly, for example, the appearance of class 5 I / O connector is the most common one See for example. If we avoid the fuzziness of this kind of connection, it will provide an effective basis for connector classification.

2.5MM Wire To Board Connectors

ShenZhen Antenk Electronics Co,Ltd , https://www.atkconnectors.com