CN113766014A

CN113766014A - Distributed anti-recommendation system and anti-recommendation method

Info

Publication number: CN113766014A
Application number: CN202110962681.XA
Authority: CN
Inventors: 张权; 刘韬; 范渊; 吴永越; 郑学新
Original assignee: Chengdu DBAPPSecurity Co Ltd
Current assignee: Chengdu DBAPPSecurity Co Ltd
Priority date: 2021-08-20
Filing date: 2021-08-20
Publication date: 2021-12-07

Abstract

The invention provides a distributed anti-recommendation system and an anti-recommendation method, which enable a user to use a system agent and randomly distribute a host to agent the access behavior of the user, enable a recommendation algorithm to be difficult to accurately acquire the access history, preference and other information of a single user, and achieve the effect of protecting the privacy of the user. Through the distributed layout, the physical hosts in the system can be distributed all over the world, and the harm of acquiring the user privacy through the geographic information is effectively avoided.

Description

Distributed anti-recommendation system and anti-recommendation method

Technical Field

The invention specifically belongs to the technical field of internet information data security, and particularly relates to a distributed anti-recommendation system and an anti-recommendation method.

Background

With the development of network technology, more and more people use networks to complete various services, such as shopping, reading news, and the like. While using these services provided by the manufacturer, the user's browsing history, preferences, geographical location, etc. are left exposed to the manufacturer. And the manufacturer accurately portrays the user according to the information of the user, and realizes personalized push. As for the user, the received push is more and more fit with the appetite, but the user is trapped in the information cocoon room. At the same time, the privacy of the user is also completely exposed to the vendor.

Privacy disclosure is one of hidden dangers brought by the current personalized recommendation algorithm, and many internet service manufacturers acquire personal information of users while providing services. There are also some anti-tracking privacy protection tools on the market that are partly based on stand-alone, but the basic principle is to limit the tracker by detecting the tracker on the website. But because the manufacturer providing the service can accurately acquire the browsing record of the user on the website providing the service, the manufacturer can still accurately acquire the favorite of the user. The scheme provides a distributed anti-recommendation algorithm, and a user accesses the service of a service manufacturer through an agent provided by the system. When a user starts to access a service provided by a manufacturer, the service is actually forwarded through a random node in the system, and each time the service is accessed, the service is realized through a different node. For a node in the system, the node receives access requests from a plurality of clients, and sends the requests from the plurality of clients to a server provided with services by a manufacturer in a time division manner. For the vendor's server to accept access from a node in the system, rather than from a single user, the vendor's recommendation algorithm becomes inefficient at this point and the user's privacy is harder to gain.

The method described in patent application with patent application No. CN201610809458.0, entitled proxy load balancing method based on mobile proxy system, has the following disadvantages:

1. anti-trace plug-in defects:

1) the plug-in is configured on the local browser, and only part of the browsing history can be prevented from being acquired. And may result in partial functionality being unavailable on a portion of the web site;

2) only aiming at the browser design, the method has no effect on APP and the like of a mobile terminal.

2. Proxy through a single node:

1) with proxy access, the vendor server can only be prevented from acquiring real IP or real geographic information. If the same node is always used for access, the browsing history and the interest can be recorded.

Disclosure of Invention

Aiming at the defects and requirements in the prior art, the invention provides a distributed anti-recommendation system and an anti-recommendation method, so that a user can use a system agent to randomly distribute a host to agent the access behavior of the user, the recommendation algorithm is difficult to accurately acquire the access history, preference and other information of a single user, and the effect of protecting the privacy of the user is achieved. Through the distributed layout, the physical hosts in the system can be distributed all over the world, and the harm of acquiring the user privacy through the geographic information is effectively avoided.

The specific implementation content of the invention is as follows:

the invention provides a distributed anti-recommendation system, which is used for receiving a real response of a manufacturer server, converting the real response into an agent response and sending the agent response to a client, and also used for receiving a real request of the client, converting the real request into an agent request and sending the agent request to the manufacturer server; a plurality of nodes are arranged in the anti-recommendation system, and a control system for selecting the nodes and a computing system for performing data interaction with a client and a manufacturer server are arranged in each node;

the computing system and the control system are connected with the client and the manufacturer server respectively.

A distributed anti-recommendation method is based on the distributed anti-recommendation system, and is used for receiving a real response of a manufacturer server, converting the real response into an agent response and sending the agent response to a client, and also used for receiving a real request of the client, converting the real request into an agent request and sending the agent request to the manufacturer server; the anti-recommendation method specifically comprises the following steps:

step 1: calculating the state of each node;

step 2: according to the states of the nodes, randomly selecting one node with a better state as a master node and the other nodes as slave nodes;

and step 3: receiving a real request of a manufacturer server, sending the real request to a main node, and distributing the received real request through a control system;

and 4, step 4: processing the real request by using a computing system, converting the real request into an agent request and then sending the agent request to a manufacturer server;

and 5: the manufacturer server makes a real response after receiving the agent request;

step 6: receiving a real response sent by a manufacturer server to a main node, and distributing the received real response through a control system of the main node;

and 7: and processing the real response by using the computing system, converting the real response into a proxy response, and sending the proxy response to the client.

In order to better implement the present invention, further, the state information of the node specifically includes whether the node is alive or not and the usage rate of the node.

In order to better implement the present invention, further, after the state of the master node is considered, the method for randomly selecting is specifically a sequential random k-algorithm.

In order to better implement the present invention, further, the selection of the node in step 2 adopts a consistent hash algorithm to select the master node.

In order to better implement the present invention, further, in step 4, when the real request is converted, a white list and a black list are set, and the specified domain name and IP access are released and intercepted.

In order to better implement the present invention, further, when the step 7 converts the real response, a filtering process is further performed: and temporarily storing the real response, filtering, and sending the filtered real response to the client.

Compared with the prior art, the invention has the following advantages and beneficial effects:

by means of the distributed system architecture, random node proxy is carried out on requests sent by users, the purposes of confusing user history records and preferences are achieved, and the effect of effectively countering recommendation algorithms of service providers is achieved. Nodes distributed throughout the locality may prevent acquisition of geo-location related parameters, as compared to existing anti-tracking plug-ins. The high-availability system formed by multiple nodes has better performance and achieves load balance. Meanwhile, a system manager can configure a white list and a black list to limit user access.

Drawings

FIG. 1 is a schematic diagram of a general framework of the system of the present invention;

FIG. 2 is a schematic flow chart of the present invention;

FIG. 3 is a schematic diagram of the present invention using a consistent hashing algorithm 1;

FIG. 4 is a schematic diagram of the present invention employing a consistent hashing algorithm, FIG. 2;

FIG. 5 is a schematic diagram of the present invention using a consistent hashing algorithm 3;

FIG. 6 is a diagram 4 illustrating the consistent hashing algorithm used in the present invention.

Detailed Description

In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it should be understood that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments, and therefore should not be considered as a limitation to the scope of protection. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, are within the scope of the present invention.

In the description of the present invention, it is to be noted that, unless otherwise explicitly specified or limited, the terms "disposed," "connected," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood in specific cases to those skilled in the art.

Example 1:

the embodiment provides a distributed anti-recommendation system, as shown in fig. 1, configured to receive a real response of a vendor server, convert the real response into an agent response, and send the agent response to a client, and also configured to receive a real request of the client, convert the real request into an agent request, and send the agent request to the vendor server; a plurality of nodes are arranged in the anti-recommendation system, and a control system for selecting the nodes and a computing system for performing data interaction with a client and a manufacturer server are arranged in each node;

The working principle is as follows: by means of the distributed system architecture, random node proxy is carried out on requests sent by users, the purposes of confusing user history records and preferences are achieved, and the effect of effectively countering recommendation algorithms of service providers is achieved. Nodes distributed throughout the locality may prevent acquisition of geo-location related parameters, as compared to existing anti-tracking plug-ins. The high-availability system formed by multiple nodes has better performance and achieves load balance. Meanwhile, a system manager can configure a white list and a black list to limit user access.

Example 2:

the embodiment provides a distributed anti-recommendation method, as shown in fig. 1 and fig. 2, based on the above-mentioned distributed anti-recommendation system, the distributed anti-recommendation system is configured to receive a real response of a vendor server, convert the real response into an agent response, and send the agent response to a client, and also configured to receive a real request of the client, convert the real request into an agent request, and send the agent request to the vendor server; the anti-recommendation method specifically comprises the following steps:

step 1: calculating the state of each node;

The working principle is as follows: the whole distributed anti-recommendation algorithm system is composed of a plurality of nodes, and the nodes are integrated for the client. The whole system consists of a plurality of nodes, and the nodes consist of a control system and a computing system. For a request, the node control system first determines which node the request is specifically proxied to according to the specific conditions of the nodes in the system, such as whether to survive, resource utilization rate, etc. For the nodes of the system, the nodes communicate with each other to determine the state of the nodes, the control system is used to distribute requests, and the computing system is used to process the requests. The configuration item description is shown in the following table 1:

TABLE 1

1. Each node in the system has a control system, but a main node exists, and the main node preferentially distributes the request. And when the main node is down and disconnected, selecting one node as the main node from the rest nodes.

2. Node selection is achieved through a sequential random k-algorithm, and other algorithms can be replaced. The method strives to achieve random selection, can achieve the effect of load balancing, and effectively avoids overlarge load of a certain node.

3. When a node processes a request, the request may be processed. If the administrator can set a white list and a black list, the access of the specified domain name and IP can be intercepted. The returned request can also be filtered, for example, the downloaded file can be temporarily stored and scanned for viruses.

Example 3:

on the basis of the above embodiment 2, it is a common matter for a distributed multi-node system to have a node failure and a disconnection. In such a case, it is desirable that only the degradation of the system performance will occur, and that no complete unavailability of the system will occur. The consistent hash algorithm is a special hash algorithm which can change the mapping relationship between the existing service request and the processing request server as little as possible when the server is moved or added.

The other parts of this embodiment are the same as those of embodiment 2, and thus are not described again.

Example 4:

this embodiment is based on the

above embodiment

2 or 3, and 1. the consistent hash algorithm is implemented by a data structure called a consistent hash ring. The start point of the ring is 0, the end point is 2^32-1, and the start point is connected with the end point, so the integer distribution range of the ring is [0, 2^32-1], as shown in FIG. 3;

suppose we have four requests that need agents, namely "Target1", "Target2", "Target3" and "Target4", the IP corresponding to the four agent requests are IP _1, IP _2, IP _3 and IP _4 respectively, and then hash values of the IP corresponding to the four agent requests are calculated by using a hash function, the range of the values is [0, 2^32-1], and the hash values are placed on a ring, as shown in fig. 4 specifically;

placing nodes (servers) onto a hash ring; the same hash function is used, the IP of the node (server) is used as a key value to carry out hash calculation, and the node position is determined according to the value. Suppose we have three nodes with corresponding IPs IP _ S1, IP _ S2, and IP _ S3.

After the agent request and the hash of the node are placed on the same hash ring, the node closest to the agent request is searched on the hash ring according to the clockwise direction, namely the node processing the agent request. Taking k3 as an example, the node closest in the clockwise direction is s 1. As shown in particular in fig. 5.

When servers are added, if the newly added node is hashed and falls between s4 and k1, only the proxy request k3 is affected, and k3 is processed by s 4.

When the servers decrease (relative to the initial state), such as after the s1 node drops, then the requests for k3, k1 are forwarded to s2 for processing.

For the newly added node, for example, the newly added s4 only shares the load of s1, and does not share the loads of other nodes, so that a virtual node is introduced. By virtualizing physically present nodes as a set of virtual nodes, e.g., one physical node is virtualized as two virtual nodes. The three physical nodes may be virtualized into six nodes. And adding the six nodes into a hash ring after hashing again, and selecting the physical node by selecting the nearest virtual node through the agent request. As shown in fig. 6.

Other parts of this embodiment are the same as any of embodiments 1 to 3, and thus are not described again.

The above description is only a preferred embodiment of the present invention, and is not intended to limit the present invention in any way, and all simple modifications and equivalent variations of the above embodiments according to the technical spirit of the present invention are included in the scope of the present invention.

Claims

1. A distributed anti-recommendation system is used for receiving a real response of a manufacturer server, converting the real response into an agent response and sending the agent response to a client, and also used for receiving a real request of the client, converting the real request into an agent request and sending the agent request to the manufacturer server; the anti-recommendation system is characterized in that a plurality of nodes are arranged in the anti-recommendation system, and each node is provided with a control system for selecting the node and a computing system for performing data interaction with a client and a manufacturer server;

2. A distributed anti-recommendation method, based on the distributed anti-recommendation system of claim 1, for receiving a real response from a vendor server, converting the real response into an agent response, and sending the agent response to a client, and for receiving a real request from the client, converting the real request into an agent request, and sending the agent request to the vendor server; the method is characterized by comprising the following steps:

step 1: calculating the state of each node;

3. The distributed counter-recommendation method of claim 2, wherein the status information of the nodes specifically includes whether the nodes survive and the usage rate of the nodes.

4. A distributed counter-recommendation method according to claim 2, wherein the method of random selection after taking into account the status of the master node is specifically a sequential random k-algorithm.

5. The distributed counter-recommendation method of claim 2, wherein the selection of the node in step 2 uses a consistent hashing algorithm to perform the selection of the master node.

6. The distributed anti-recommendation method of claim 2, wherein step 4 sets a white list and a black list for releasing and intercepting the specified domain name and IP access when converting the real request.

7. A distributed anti-recommendation method according to claim 2, wherein said step 7 is further performed with a filtering process when converting the real response: and temporarily storing the real response, filtering, and sending the filtered real response to the client.