Data obfuscation system and data obfuscation method
Through the obfuscation device and balance generator in the data obfuscation system, the network address is replaced and generated, and the confidential data leakage caused by inquiring confidential problems from external artificial intelligence websites is solved, and data security and cost-effectiveness are improved.
Patent Information
- Application Number
- CN202410084285.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-01-19
- Publication Date
- 2025-07-22
AI Technical Summary
In the prior art, users asking confidential questions from external artificial intelligence websites may lead to leaks of confidential data, and establishing an artificial intelligence website on their own is time-consuming and costly.
The data obfuscation system is adopted, which includes obfuscation devices, network address generators and balance generators. By replacing target words, generating multiple network addresses and input errors, the automatic learning mechanism of artificial intelligence websites is offset and confidential data leakage is avoided.
Effectively avoiding confidential data leakage, saving time and cost of building artificial intelligence websites, and improving data security.
Smart Images
Figure CN120354385A_ABST
Abstract
Description
Technical Field
[0001] The present disclosure relates to a data obfuscation system and a data obfuscation method, and particularly to a data obfuscation system and a data obfuscation method that perform data obfuscation and counteract the automatic learning mechanism of an artificial intelligence website through the collaborative operation of an obfuscation device and a balance generator. Background Art
[0002] With the progress of technology, artificial intelligence websites can assist in answering users' questions, enabling users to obtain answers quickly and accurately. However, if a user asks some confidential questions to an external artificial intelligence website, it may lead to the leakage of confidential data. If the user builds an artificial intelligence website by themselves, it not only takes a huge amount of time but also has a high cost. Summary of the Invention
[0003] In view of the deficiencies of the prior art, one of the objectives of the present disclosure is (but not limited to) to provide a data obfuscation system and a data obfuscation method to improve the deficiencies of the prior art.
[0004] In some embodiments, the data obfuscation system includes an obfuscation device, a network address generator, and a balance generator. The obfuscation device is used to replace at least one target word of the input data with at least one preset word to generate output data, and transmit the output data to the artificial intelligence website. The network address generator is used to generate a plurality of network addresses. The balance generator is used to randomly input at least one error to at least one target code of the input data or modify at least one target code of the input data to generate a plurality of corresponding data. The balance generator transmits the first corresponding data and the second corresponding data of the corresponding data to the artificial intelligence website through the first network address and the second network address of the network addresses respectively. The first corresponding data is related to the input data, and the second corresponding data is not related to the input data.
[0005] In some embodiments, the data obfuscation method includes: replacing at least one target word of the input data with at least one preset word through an obfuscation device to generate output data, and transmitting the output data to the artificial intelligence website; generating a plurality of network addresses through a network address generator; randomly inputting at least one error to at least one target code of the input data or modifying at least one target code of the input data through a balance generator to generate a plurality of corresponding data; and transmitting the first corresponding data and the second corresponding data of the corresponding data to the artificial intelligence website through the first network address and the second network address of the network addresses respectively by the balance generator. The first corresponding data is related to the input data, and the second corresponding data is not related to the input data.
[0006] The technical means embodied in the embodiments of the present disclosure can improve at least one of the disadvantages of the prior art. The data obfuscation system and data obfuscation method of the present disclosure can perform data obfuscation through the coordinated operation of an obfuscation device and a balance generator, thereby effectively solving the situation of confidential data leakage caused by asking confidential questions to external artificial intelligence websites. In addition, since the data obfuscation system and data obfuscation method of the present disclosure can effectively avoid the leakage of confidential data, users do not need to build their own artificial intelligence websites to avoid data leakage. Thus, the present disclosure can not only save the time for building an artificial intelligence website, but also save the huge cost of building an artificial intelligence website.
[0007] Regarding the features, implementation, and technical effects of the present disclosure, the preferred embodiments will be described in detail below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0008] Figure 1 FIG. 1 is a schematic diagram of a data obfuscation system drawn according to some embodiments of the present disclosure; and
[0009] Figure 2 FIG. 2 is a flowchart of a data obfuscation method drawn according to some embodiments of the present disclosure.
[0010] SYMBOL DESCRIPTION
[0011] 100: Data obfuscation system
[0012] 110: Obfuscation device
[0013] 111: Obfuscation server
[0014] 113: De-obfuscation server
[0015] 120: Network address generator
[0016] 130: Balance generator
[0017] 140: Timer
[0018] 150: Database
[0019] 200: Method
[0020] 210-250: Steps
[0021] 700: Electronic device
[0022] 710: Firewall
[0023] 800: Internet
[0024] 900: Artificial intelligence website DETAILED DESCRIPTION OF THE EMBODIMENTS
[0025] All terms used in this document have their ordinary meanings. The definitions of the above terms in commonly used dictionaries are only examples of the use of any term discussed herein and should not limit the scope and meaning of this disclosure. Similarly, this disclosure is not limited only to the various embodiments shown in this specification.
[0026] Regarding the use of "coupled" or "connected" herein, it can refer to two or more elements being in direct physical or electrical contact with each other, or being in indirect physical or electrical contact with each other, and can also refer to two or more elements operating or acting on each other. As used herein, the term "circuit" can be a device in which at least one transistor and / or at least one active or passive element are connected in a certain manner to process signals.
[0027] As used herein, the term "and / or" includes any combination of one or more of the listed related items. In this document, the use of terms such as first, second, and third is for describing and distinguishing each element. Therefore, the first element in this document can also be referred to as the second element without departing from the meaning of this disclosure. For ease of understanding, similar elements in each drawing will be designated with the same reference numerals.
[0028] To improve the situation in the prior art where asking some confidential questions to external artificial intelligence websites may lead to the leakage of confidential data, this disclosure proposes a data obfuscation system and a data obfuscation method, which are described in detail as follows.
[0029] Figure 1 FIG. is a schematic diagram of a data obfuscation system 100 drawn according to some embodiments of this disclosure. As shown in the figure, the data obfuscation system 100 includes an obfuscation device 110, a network address generator 120, and a balance generator 130. To facilitate the understanding of the operation of the data obfuscation system 100, please also refer to Figure 2 , Figure 2 FIG. is a flowchart of a data obfuscation method 200 drawn according to some embodiments of this disclosure.
[0030] Please refer to Figure 1 and Figure 2 , in step 210, at least one target word of the input data is replaced with at least one preset word by the obfuscation device 110 to generate output data, and the output data is transmitted to the artificial intelligence website.
[0031] For example, when a user wants to ask a question to the artificial intelligence website 900, the user can input the question to be asked through the electronic device 700. To prevent the leakage of confidential data, the firewall 710 of the electronic device 700 can store the Internet Protocol addresses (IP addresses) of multiple different artificial intelligence websites 900, such as storing the Internet Protocol address of Chat Generative Pre-trained Transformer (ChatGPT). Once the firewall 710 detects that the electronic device 700 wants to connect to the artificial intelligence website 900, the firewall 710 will direct the question of the electronic device 700 to the data obfuscation system 100. Conversely, the firewall 710 will direct the question of the electronic device 700 to the Internet 800.
[0032] Next, when the obfuscation device 110 receives the question from the electronic device 700, the obfuscation device 110 can replace the target word (such as Company A) of the question with a preset word (such as Company B) to generate a modified question, and transmit the modified question to the artificial intelligence website 900. In this way, the present disclosure can replace the confidential data (such as Company A) with other data, thereby effectively solving the situation where the user asks a confidential question to an external artificial intelligence website 900 and causes the leakage of confidential data. It should be noted that the present disclosure is not limited to the above embodiments, and the user can replace different keyword words (such as specific models, customer names, etc.) with other words, depending on the actual needs.
[0033] In some embodiments, the obfuscation device 110 includes a de-obfuscation server 113, and the de-obfuscation server 113 is used to receive the response data corresponding to the output data from the artificial intelligence website 900, and restore at least one preset word in the response data to at least one target word to generate result data. For example, the de-obfuscation server 113 can receive the answer to the question from the artificial intelligence website 900, and restore the preset word (such as Company B) in the answer to the target word (such as Company A) to generate a real result for the user to reference.
[0034] In some embodiments, the data obfuscation system 100 further includes a timer 140 that provides a preset time. The obfuscation device 110 further includes an obfuscation server 111 that is configured to transmit output data to the artificial intelligence website 900 at a first time point, and the de-obfuscation server 113 receives response data from the artificial intelligence website 900 at a second time point. For example, the timer 140 uses 60 seconds as the preset time. The obfuscation server 111 transmits a question to the artificial intelligence website 900 at 9:00:00 am, and the de-obfuscation server 113 receives a reply from the artificial intelligence website 900 at 9:00:59 am. The time difference between the two is 59 seconds, which is less than the preset time of 60 seconds. Therefore, the de-obfuscation server 113 restores the preset word (such as Company B) in the reply to the target word (such as Company A) to generate a real result for the user's reference.
[0035] In some embodiments, if the time difference between the first time point and the second time point is greater than the preset time, the de-obfuscation server 113 does not perform a de-obfuscation operation on the response data. For example, continuing with the above embodiment, the timer 140 uses 60 seconds as the preset time. The obfuscation server 111 transmits a question to the artificial intelligence website 900 at 9:00:00 am, and the de-obfuscation server 113 receives a reply from the artificial intelligence website 900 at 9:01:15 am. The time difference between the two is 1 minute and 15 seconds, which is greater than the preset time of 60 seconds. Therefore, the de-obfuscation server 113 does not de-obfuscate the reply to obtain a result.
[0036] In step 220, a plurality of network addresses are generated by the network address generator 120.
[0037] For example, the network address generator 120 can generate a plurality of Internet Protocol addresses (IP addresses). When the obfuscation device 110 receives a question from the electronic device 700, the obfuscation device 110 can replace the target word (such as Company A) of the question with a preset word (such as Company B), and also change the Internet Protocol address (IP address), and transmit the modified question to the artificial intelligence website 900. In this way, in addition to obfuscating the real question content, the present disclosure also obfuscates the Internet Protocol address (IP address) from which the question is sent, thereby more effectively solving the situation where the user asks a confidential question to an external artificial intelligence website 900 and causes the leakage of confidential data, and even avoiding the leakage of the data of the inquirer of the confidential data.
[0038] In step 230, at least one target encoding of the input data is randomly input with at least one error by the balance generator 130 or at least one target encoding of the input data is modified to generate a plurality of corresponding data. For example, the balance generator 130 may randomly input an error (such as "error") to the target encoding of the question (such as "code") or modify the target encoding of the question (such as "code") to generate a plurality of modified encodings. After receiving the response from the artificial intelligence website 900, the randomly input error ("error") is removed, or the modified target encoding (such as "code") is restored to generate a real result for the user to refer to.
[0039] In step 240, the balance generator 130 transmits the first corresponding data and the second corresponding data of the corresponding data to the artificial intelligence website through the first network address and the second network address of the network addresses respectively. The first corresponding data is related to the input data, and the second corresponding data is not related to the input data. For example, to offset the automatic learning mechanism of the artificial intelligence website 900, the balance generator 130 may generate multiple sets of similar questions and ask the artificial intelligence website 900 at different network protocol addresses and time points, so that the learning effects of the artificial intelligence website 900 cancel each other out.
[0040] In some embodiments, the balance generator 130 receives the first response data corresponding to the first corresponding data from the artificial intelligence website 900, and receives the second response data corresponding to the second corresponding data from the artificial intelligence website 900, and generates result data according to the first response data corresponding to the first corresponding data. For example, continuing with the above embodiment, among multiple sets of similar questions, at least one set of similar questions is related to the real question, and the remaining similar questions are not related to the real question. The remaining similar questions are only proposed to cancel out the learning effects of the artificial intelligence website 900. Therefore, even if the balance generator 130 receives the responses to all similar questions from the artificial intelligence website 900, the balance generator 130 will only generate an inquiry result according to the real question for the user to refer to.
[0041] In some embodiments, the balance generator 130 also transmits the corresponding data to the artificial intelligence website 900 through the network addresses, and the number of the corresponding data transmitted by the balance generator 130 is less than a preset threshold. For example, if the balance generator 130 sends too many questions to the artificial intelligence website 900, it may be considered a malicious attack by a hacker by the artificial intelligence website 900 and thus be blocked by the artificial intelligence website 900. Therefore, the present disclosure sets a preset threshold to limit the number of questions sent by the balance generator 130 to avoid being blocked by the artificial intelligence website 900. In some embodiments, the preset threshold may be 100 times, but the present disclosure is not limited to this embodiment. The user can set the preset threshold to 10 times or 1000 times according to different situations, depending on the actual needs.
[0042] In some embodiments, the data obfuscation system 100 further includes a database 150. The database 150 is used to provide at least one preset word corresponding to at least one target word. For example, the database 150 stores a preset word (such as Company B). When the obfuscation device 110 receives a question from the electronic device 700, the obfuscation device 110 can replace the target word (such as Company A) of the question with the preset word (such as Company B) stored in the database 150 to generate a modified question.
[0043] In some embodiments, the number of at least one preset word is multiple, and the database 150 provides one of the preset words corresponding to at least one target word according to different time points. For example, the preset words provided by the database 150 change over time. For example, the preset word provided by the database 150 at 9:00:00 am is Company B. When the obfuscation device 110 receives a question from the electronic device 700, the obfuscation device 110 can replace the target word (such as Company A) of the question with the preset word (such as Company B) provided by the database 150 to generate a modified question.
[0044] In addition, the preset word provided by the database 150 at 9:01:15 am is changed to Company C. When the obfuscation device 110 receives a question from the electronic device 700, the obfuscation device 110 can replace the target word (such as Company A) of the question with the preset word (such as Company C) provided by the database 150 to generate a modified question. Subsequently, the modified question is sent to the artificial intelligence website 900. After receiving the reply from the artificial intelligence website 900, the preset word (such as Company B or Company C) is substituted back with the target word (such as Company A). Since the present disclosure can change the keyword to different words over time, at different time points, Company A will be replaced with Company B or Company C, and only the present disclosure knows which word the keyword is replaced with. Thus, the risk of confidential data leakage can be further reduced.
[0045] In some embodiments, the obfuscation device 110 further replaces the target model of the input data with a preset model. For example, when the obfuscation device 110 performs the obfuscation operation on the Log file, the present disclosure can replace the model with other models to avoid the leakage of confidential data.
[0046] In some embodiments, the obfuscation device 110 further replaces the target image of the input data with a preset image. For example, the obfuscation device 110 can replace the trademark image of Company A with the trademark image of Company B to avoid the leakage of confidential data. It should be noted that the present disclosure is not limited to the above embodiments. The obfuscation device 110 can replace different images (such as portrait, building image... etc.) with other images depending on actual needs.
[0047] It should be noted that this disclosure is not limited to Figures 1 to 2 the embodiments shown, which are only used to exemplarily show one of the implementation manners of this disclosure, so as to make the technology of this disclosure easy to understand. The claims of this disclosure shall be subject to the invention claims. Those skilled in the art, without departing from the concept of this disclosure, the modifications and refinements made to the embodiments of this disclosure still fall within the invention claims of this disclosure.
[0048] In summary, the data obfuscation system 100 and the data obfuscation method 200 of this disclosure can perform data obfuscation through the collaborative operation of the obfuscation device 110 and the balance generator 130, thereby effectively solving the situation of confidential data leakage caused by asking confidential questions to the external artificial intelligence website 900. In addition, since the data obfuscation system 100 and the data obfuscation method 200 of this disclosure can effectively avoid the leakage of confidential data, users do not need to build their own artificial intelligence websites to avoid data leakage. Thus, this disclosure can not only save the time for building an artificial intelligence website, but also save huge costs for building an artificial intelligence website.
[0049] Although the embodiments of this disclosure are as described above, these embodiments are not used to limit this disclosure. Those skilled in the art of this technology can make changes to the technical features of this disclosure according to the explicit or implicit content of this disclosure. All such changes may fall within the scope of patent protection sought by this disclosure. In other words, the scope of patent protection of this disclosure shall be subject to what is defined by the claims of this specification.
Claims
1. A data obfuscation system, comprising: An obfuscation device for replacing at least one target word of an input data with at least one preset word to generate an output data, and transmitting the output data to an artificial intelligence website; A network address generator for generating a plurality of network addresses; And A balance generator for randomly inputting at least one error to at least one target code of the input data or for modifying the at least one target code of the input data to generate a plurality of corresponding data, wherein the balance generator transmits a first corresponding data and a second corresponding data of the plurality of corresponding data to the artificial intelligence website using a first network address and a second network address of the plurality of network addresses respectively, wherein the first corresponding data is related to the input data, and the second corresponding data is not related to the input data.
2. The data obfuscation system according to claim 1, wherein the obfuscation device comprises: A de-obfuscation server for receiving a response data corresponding to the output data from the artificial intelligence website, and restoring the at least one preset word of the response data to the at least one target word to generate a result data.
3. The data obfuscation system according to claim 2, further comprising: A timer for providing a preset time; Wherein the obfuscation device further comprises: An obfuscation server for transmitting the output data to the artificial intelligence website at a first time point, and the de-obfuscation server receiving the response data from the artificial intelligence website at a second time point, Wherein if a time difference between the first time point and the second time point is less than or equal to the preset time, the de-obfuscation server restores the at least one preset word of the response data to the at least one target word to generate the result data.
4. The data obfuscation system according to claim 3, wherein if the time difference between the first time point and the second time point is greater than the preset time, the de-obfuscation server does not perform a de-obfuscation operation on the response data.
5. The data obfuscation system according to claim 1, wherein the balance generator receives a first response data corresponding to the first corresponding data from the artificial intelligence website, and receives a second response data corresponding to the second corresponding data from the artificial intelligence website, and generates a result data according to the first response data corresponding to the first corresponding data.
6. The data obfuscation system according to claim 1, wherein the balance generator further transmits the plurality of corresponding data to the artificial intelligence website using the plurality of network addresses, wherein a quantity of the plurality of corresponding data transmitted by the balance generator is less than a preset threshold.
7. The data obfuscation system according to claim 1, further comprising: A database for providing the at least one preset word corresponding to the at least one target word.
8. The data obfuscation system according to claim 7, wherein a quantity of the at least one preset word is multiple, and the database provides one of the multiple preset words corresponding to the at least one target word according to different time points.
9. A data obfuscation method, comprising: At least one target word of an input data is replaced with at least one preset word by a confusion device to generate an output data, and the output data is transmitted to an artificial intelligence website; A plurality of network addresses are generated by a network address generator; At least one error is randomly input to at least one target code of the input data or the at least one target code of the input data is modified by a balance generator to generate a plurality of corresponding data; and The balance generator transmits a first corresponding data and a second corresponding data of the plurality of corresponding data to the artificial intelligence website through a first network address and a second network address of the plurality of network addresses, wherein the first corresponding data is related to the input data, and the second corresponding data is not related to the input data.
10. The data confusion method according to claim 9, further comprising: Receiving, by a deconfusion server, a response data corresponding to the output data from the artificial intelligence website, and restoring the at least one preset word of the response data to the at least one target word to generate a result data.