TWI433037B - Method for user intention-awareness, electronic device and data processing system therefor - Google Patents

Description

Method, electronic device and data computing system for sensing user intention

The present invention relates to techniques for perceiving user intent, and more particularly to a method, an electronic device, and a data computing system for sensing user intent.

The development of environment-aware technology provides users with a series of more personalized services. In such an environment, how to help users quickly search for the services they really need in different perception environments is close to the needs of users. This is a complex issue.

In response to the needs of this issue, how to establish a model for identifying user behaviors, so as to predict the user's intentions, so that the interaction between the individual and the use of environmental information is good, and the user's basic behavior and information are used to establish sensing. The User Profile of the user's intent, directly providing predictions of user services, is one of the simplest solutions.

In addition, in the current operating environment, the massive use of handheld devices has generated a large amount of time and space data. To use the handheld device as a tool for providing or acquiring personalized services, and to predict the user's intentions, it is necessary to use the data effectively. Therefore, how to deal with these materials, and to establish appropriate user files, and to respond to the situation of individual users, to make appropriate services, has become an important issue in the perception of environmental service technology.

The invention relates to an intent sensing method, which establishes a user keyword network according to digital content browsed by a user and at least one knowledge ontology, presents the content according to the structured content, and structurally categorizes the content according to the user attribute. The need to create structured content with inferred user intent. Since the requirements related to user attributes can be combined with time, location or other parameters, the above structured content with inferred user intent can be used as a basis for providing personalized services to achieve the needs of environment-aware technology.

According to an aspect of the present invention, an intent sensing method is provided, comprising the following steps. A user keyword network is constructed according to the content browsed by the user on an electronic device and at least one knowledge ontology. According to a service requirement, the data content related to the service demand in the user keyword network is extracted through the mutation analysis, and the user intention inference is performed to generate the user intention to perceive the content in an intention-aware profile file.

According to another aspect of the present invention, an electronic device having a user's intention is provided, including: a user interface unit and a processing module. User interface unit for interactive presentation of data. The processing module is configured to construct a user keyword network according to the content browsed by the user interface unit and the at least one knowledge ontology. The processing module responds to a service requirement, and extracts the data content related to the service requirement in the user keyword network through the mutation analysis, and performs user intention inference to generate the user intention to perceive the content.

According to still another aspect of the present invention, a data computing system having a user intent is provided, comprising: a communication unit and a processing unit. The communication unit is configured to receive content and events browsed by one of the users. The processing unit is configured to construct a user keyword network according to the content and events browsed by the received user and the at least one knowledge ontology. The processing unit responds to a service requirement, and extracts the data content related to the service requirement in the user keyword network through the mutation analysis, and performs user intention inference to generate the user intention to perceive the content.

In order to make the above-mentioned contents of the present invention more comprehensible, the preferred embodiments are described below, and the detailed description is as follows:

According to the present invention, a method, an electronic device and a system for user intent perception are proposed. Please refer to FIG. 1A, which is an embodiment of a method for sensing a user's intention according to the present invention. As shown in FIG. 1A, in step S100, a user keyword network 30 is constructed according to the content 20 browsed by the user on an electronic device and the at least one knowledge body 10. Step S110, according to a service requirement, extract the content of the user keyword network 30 related to the service demand through the mutation analysis, and perform user intention inference to generate the user intention to perceive the content in an intention-aware profile file 40. .

As shown in FIG. 1A, step S110 includes, for example, steps S120 and S130. In step S120, according to the service requirement, the content of the variance analysis is generated by intercepting the data related to the service demand from the user keyword network 30. In step S130, an association between the keywords of the content of the variation analysis is intentionally inferred to generate a user intention to perceive the content.

Moreover, for the ontology 10, embodiments may be statically obtained or implemented to be dynamically generated or updated. The user keyword network 30, for example, includes structured content of a plurality of nodes, each node corresponding to one or more keywords and corresponding weight values. In an embodiment, the requirements of the implementation or application are more visual, such that the keyword is associated with a plurality of attributes, such as time or spatial parameters, via the browsing content from which the keyword originated. The user's intention to perceive the content contained in the intent-aware profile 40 may, in one embodiment, also represent the keywords and their associations. Other examples will be explained later.

The above method embodiments have various implementations, such as implementing the method of FIG. 1A as a software. Whether implemented as a system, application software or program mode or firmware, it can be installed and applied in an electronic device 200 as shown in FIG. 2A, such as an arithmetic device such as a computer, or a mobile device such as a mobile phone or multimedia. Players, personal digital assistants, navigation devices, laptops, and car computers. This embodiment uses the time and space data generated by the mobile device and based on the content browsed by the user, and the generated user side file can be used as a basis for providing personalized service to achieve the environment-aware technology. Demand.

The electronic device 200 basically includes a user interface unit 210, such as a display, an input device, or other input or output device that interacts with the use, such as a touch screen, and the user browses the content from the user interface unit 200, such as online information. Or the content of the website and its browsing behavior, such as clicking a link, selecting a content, or searching for text, may be used as a basis for constructing the user keyword network 30 shown in FIG. 1A. In addition, the processing module 212 of the electronic device 200 executes software that implements the user's intention to be perceived according to the above method.

For example, the electronic device 200 is a function of a mobile phone having a positioning function such as a location-based services (LBS) or a global positioning system (GPS). If the system software detects that the time is already 6:00 pm, a service request is made to provide the user with a suggested meal message, and the user intention sensing software executed by the processing module 212 includes the time condition and dinner. The service request and the location of the region where the current user is located, from the user keyword network 30 in accordance with the intent to sense the profile 40 generated or updated in step S110, is obtained, for example, in the vicinity of the user's area, and is in accordance with the user. Relevant restaurant name, taste, and dish color. The system software can use these inferences in response to the above-mentioned service requirements (including parameters such as keywords or time or location) to be processed to provide different personalized services or to make the phone more intelligent.

Another practical example is that the user having the sensory intention in accordance with the above method embodiment is installed in the data computing system, as shown in FIG. 2B, the servo system 270. Basically, the servo system 270 has the communication unit 271 and different clients. The end devices, such as the electronic devices 201, 202, communicate via a communication link 250 in a wireless or wired communication manner (such as a broadband network, a wireless local area network, or a 3G, 3.5G mobile communication link). Assuming that the user wants to use the intent awareness function, the user can log in to the servo system 270 to allow the browsed content to be passed to the servo system 270, wherein the content being viewed, for example, non-structural content such as natural language, or Structured content such as data content with certain established fields or items. The processing unit 273 of the servo system 270 (such as a system formed by a computer or a plurality of servers) executes a software having a perceived user's intention, which may be generated by the client device or the servo system according to step S110 or The updated intent-aware profile 40 can be used as a basis for providing personalized services, such as being provided to a third service provider other than the servo system 250 or by the servo system 250 itself.

Additionally, FIG. 3 illustrates a system architecture diagram of a system 300 having perceived user intent in accordance with an embodiment of the present invention. In the third diagram, the system 300 can be regarded as the software system architecture having the user's intention of being perceived in the electronic device 200 of FIG. 2A or the servo system 270 of FIG. 2B. The system 300 includes a user keyword network module 310, a mutation analysis module 320, and a user intention-aware inference module 330, which can be used to implement steps S100, S120, and S130, respectively, and generate user keyword networks respectively. 30. Variation analysis content 35 and user side write file (or intentional perception profile file) 40. The knowledge ontology 10, the user keyword network 30, the mutation analysis content 35, and the user side write file 40 may be implemented as stored in a database of the server or in the memory of the client, or in a distributed storage. Yes, if necessary, you can also use the communication link to obtain the above content and update it. In FIG. 3, the system 300 can provide the manager's knowledge ontology 10 with control functions, rights, and other control functions and rights that are open to other users.

For the ontology of knowledge, embodiments may be statically obtained from outside the system, or implemented as being dynamically generated or updated by the client or the server. Therefore, the embodiment of FIG. 3 further includes a knowledge body module 301 for generating the knowledge body 10.

The following is an example of each module in the third figure.

Knowledge ontology

The ontology module 301 can perform the steps of constructing the ontology as described above. The flowchart of Fig. 1B is an embodiment of constructing an ontology. Step S210: Perform content processing to obtain feedback according to the provided content, such as the content of the online information system, such as website content or professional content or domain knowledge content or other established knowledge ontology architecture and content. Vocabulary. Website content such as web pages, blogs, websites, and domain knowledge such as literature, technology, society, finance, politics, or life-related such as dining, transportation, sports, entertainment, etc. The content of different fields may be based on other file contents, wherein the above content may be non-structural content such as natural language, or structured content may also be implemented. The knowledge ontology module 301, as shown in step S220, uses the file exploration technology, such as classification, grouping, or other tree structure, to generate a knowledge ontology structure in which the vocabulary is a node, that is, a vocabulary network Knowledge ontology. Step S230, storing the knowledge ontology in a database or other storage memory. If the structured content is represented as a schema, then as part of the ontology as shown in Figure 4, each vocabulary may have its own weight value, indicating the frequency ratio as the vocabulary appears in the content, for example The restaurant is 0.0014. In addition, the ontology can be an unequal vocabulary network, and each node can be attached with an unlimited number of related keywords. For example, the knowledge base system is implemented to be constructed and described in an extensible markup (XML). In addition, in other examples, the existing knowledge ontology such as the SUMO (Suggested Upper Merged Ontology) of the IEEE, or the WordNet of Princeton University can also be used as a reference for the architecture of the ontology.

User keyword network

The user keyword network module 310 is based on, as shown in step S100, the user browsing behavior is based on the knowledge ontology analysis content to become the user keyword network 30. Figure 5 shows an embodiment of generating a user keyword network. Step S510, recording a user interactive browsing event with the online information system into a log, and storing the browsing content record user browsing content. Step S510 can be executed by the client by another module, such as a log and content module, and transmitted back to the server for processing; in addition, it can be implemented by the electronic device for recording and processing. It can also be processed by the user keyword network module 310 (data pre-processing part). Step S520, performing file exploration processing on the stored browsing content according to the record of the user log, for example, the action of breaking the word, and obtaining the feedback vocabulary. Step 530: Establish a user keyword network 30 according to the feedback vocabulary and the knowledge ontology. The keyword network 630, as illustrated in FIG. 6, is composed of keywords as nodes, and each keyword may have a weight value. For example, each keyword may be given a multiplication or a calculation according to the demand, resulting in different The weight value is associated as an attribute. In addition, keyword network 630 is also constructed and described in an extensible markup language. Step S540: Associate a plurality of attributes including a time parameter and a spatial parameter according to the browsing content of the keyword keyword network. For example, the keyword "fruit salad" in Figure 6 is related to the location of a webpage from a restaurant A in Hsinchu and a restaurant B in Taipei. The location of the restaurant B is known, for example, from a web page or from a web page. The geospatial server or the search website knows that X=121.5475800000, Y=25.0404700000, Z=0.00000, and the browsing time is T=2009-05-03, 07:07:18.002. This time, spatial parameter, such as a set of parameters 642 in Figure 6, is associated with the keyword "fruit salad" and other keywords such as "apple." In addition, as shown in FIG. 6, another set of parameters 641 may also be associated with a plurality of different keywords. Therefore, a keyword can have multi-dimensional attributes.

It is worth noting that the establishment of the above attributes (such as time and space parameters) and associated with the keywords may be implemented according to different purposes or methods established or used by the user. For example, the attributes such as time and space may be different depending on the generation mode or the content, such as the user usage time, the webpage time, the login system time (such as the electronic device 201 logging into the servo system 270 of FIG. 2B) or the specific time. Time definition. The location of the spatial parameters may also be based on, for example, the location of the store (such as Figure 6 above), the location of the content (such as the location of the user at the time of browsing), or other definitions. Therefore, the example in the above FIG. 6 is only an example. The embodiment of the present invention can define different time and space parameters according to different application requirements, and even add other parameters, and derive different applications. For example, you can even add variability attributes, such as attributes that change due to time, space, browsing behavior, or requirements, to produce user information with a dynamic state. As such, the embodiment of the present invention has great flexibility and extensibility, can effectively utilize the mobile device to generate a large amount of time and space data in the current action environment, and can be provided by the handheld device as a personalized service or The tools obtained to predict user intent.

In addition, referring to FIG. 2B, for the system 300 of FIG. 3, in an embodiment, the user may be provided with control functions, rights, and other functions for adding, deleting, and updating each node. Different levels of control functions and permissions.

Variation attribute analysis

In step S120, according to the service requirement, the content of the variance analysis is generated by intercepting the data related to the service demand from the user keyword network 30. Service requirements, which may include, for example, keywords, or a combination of attributes, such as those described above with respect to the user keyword network 30, such as time, fixed point, or other attributes, or even variability attributes. For example, the foregoing intercepts the content of the user's vocabulary tree structure to generate user information with dynamic properties. As shown in FIG. 7A, from the keyword network 630, in response to a service demand, if the data after June 3 is needed, the vocabulary of the relevant date is found, and the item as shown in FIG. 7A is extracted. According to this, the node merging method is used, and the rules of the tree structure are established by using frequent items, and the merging or growth calculus is performed to generate the variation analysis content 35. For example, using the FP growth tree algorithm, such as a minimum support, for example, set to 0.5, produces large items to produce, for example, a full-binary tree. Among them, according to the keyword aggregation of the node, according to the merge rule, the correlation between the nodes is calculated, and multiple branches are integrated to form a new architecture, for example, the result shown in FIG. 7B. In Fig. 7B, the portion of the tree diagram shown by the broken line 710 corresponds to the item of Fig. 7A extracted from the above-described service demand example. According to the spirit of the above-mentioned variation analysis and node merging method in data exploration, generally, knowledge can be used as various related methods in data exploration to generate variation analysis content. As such, the details will not be described again. As described above, the mutation analysis module 320 in FIG. 3 can also generate the variation analysis content 35 for the user's intention to perceive the inference module 330 in accordance with the various embodiments described above.

User inference

Inferential calculus or other algorithmic calculations are performed on the content of the variance analysis using attributes. The inference method uses, for example, the Bayesian algorithm or other intent deductive calculus to establish the correlation between keyword classes, and then produces causal inference. Using different parts of the parent calculation or different algorithm calculations will produce different results, which can be adjusted according to actual needs.

For example, the content of the variation analysis shown in Figure 7B is used to obtain the correlation between the two words. For example, the correlation between the two words of "Sushi-Avocado", "Sushi-Life", "Sushi-竘蒻", "Life--" and "Menu-Sushi" is calculated according to the maternal value of the demand. The ratio is taken to obtain the correlation between the two vocabularies. As shown in the diagram on the left side of Figure 8A and Figure, the value of the correlation between the two vocabulary (between 0 and 1) is shown on the link between the two words. Based on this, it can be deduced as shown in Figure 8A. The result of the deduction shown on the right, such as the menu associated with sushi. In addition, according to different algorithms, the different maternal values will obtain different correlation values and produce different derivation results. As shown in Fig. 8B, there is a linear correlation between avocado, sushi, vitality and sputum. In addition, FIG. 9 shows another example, using the above-described intentional inference (such as causal inference) to generate a user's intention to perceive the content.

According to the above-mentioned spirit of using the causal inference method in data exploration, usually knowledge can be inferred by the above intentions by various related methods in data exploration. As such, the details will not be described again. As described above, the user intent-aware inference module 330 in FIG. 3 can also generate an intent-aware profile file 40 that the user intends to perceive the content to generate or update in accordance with the various embodiments described above.

Furthermore, in accordance with an embodiment of the present invention, a computer or computing device (such as shown in FIG. 2A or 2B) is readable information storage medium having stored thereon a code representing one or more program modules as described above. As shown in Fig. 3, this code can be used to implement embodiments of the method of user perception of the present invention (as in the various embodiments shown in Figures 1A, 1B, and 5). The computer readable information storage medium of the embodiment is, for example but not limited to, an optical information storage medium, a magnetic information storage medium or a memory such as a memory card, a firmware or a ROM or a RAM.

The method, the electronic device, and the system for intent-awareness of the user disclosed in the above embodiments, based on the digital content browsed by the user and the at least one knowledge body, establish a user keyword network to present the structured content, and according to the structure The content, in response to the needs of the user attributes, establish structured content with inferred user intent. The intention to perceive the generation of profile files and can be continuously updated with the service requirements, so as the basis for providing personalized services, to achieve the needs of perceived environmental services.

In conclusion, the present invention has been disclosed in the above preferred embodiments, and is not intended to limit the present invention. A person skilled in the art can make various changes and modifications without departing from the spirit and scope of the invention. Therefore, the scope of the invention is defined by the scope of the appended claims.

10. . . Knowledge ontology

20. . . User browsing content

30,630. . . User keyword network

35. . . Variation analysis content

40. . . Intention-aware profile

200~202. . . Electronic device

210. . . User interface unit

212. . . Processing module

250. . . Communication link

270. . . server system

271. . . server system

273. . . Processing unit

300. . . User intent perception system

301. . . Knowledge ontology module

310. . . User keyword network module

320. . . Variation analysis module

330. . . User intent-aware inference module

641, 642. . . Vocabulary attribute

710. . . Part of the tree diagram

S100-S130, S210-S230, S510-540. . . step

Figure 1A is an illustration of one embodiment of a method for sensing a user's intent in accordance with the present invention.

FIG. 1B is a diagram of a method of constructing an ontology according to an embodiment of the present invention.

An electronic device according to an embodiment of the present invention, which can be used to sense the intention of the user, as shown in FIG. 2A.

Figure 2B illustrates a servo system that can be used to sense the user's intended side in accordance with an embodiment of the present invention.

Figure 3 illustrates a system architecture with perceptual user intent in accordance with the present invention.

Figure 4 illustrates a portion of an ontology.

Figure 5 illustrates an embodiment of generating a user keyword network.

An example of a user keyword network illustrated in FIG.

Figures 7A and 7B schematically illustrate one embodiment of utilizing mutation analysis to generate variation analysis content.

Figures 8A and 8B schematically illustrate an example of using a causal inference method to derive two different user intents to perceive content.

Figure 9 shows another example of a user's intention to perceive content using a causal inference method.

Claims (29)

A method for perceiving a user's intention includes: constructing a user keyword network according to content viewed by the user on an electronic device and at least one knowledge ontology; and extracting the usage through mutation analysis according to a service requirement The content of the information in the keyword network related to the service requirement is inferred by the user intent to generate the user's intention to perceive the content in an intention-aware profile. The method of claim 1, wherein the knowledge base system is structured content based on content provided by an online information system or content of a professional domain or an established ontology architecture and content. For example, the method described in claim 1 further comprises: constructing the knowledge ontology according to the content provided by the online information system or the content of the professional domain or the established ontology structure and content. The method of claim 3, wherein the step of constructing the knowledge ontology comprises: performing content processing to obtain feedback according to content provided by an online information system or professional domain content or an established knowledge ontology architecture and content. The vocabulary; the vocabulary of the feedback, the file exploration operation to establish a vocabulary network as the knowledge ontology. The method of claim 4, wherein the method further comprises: storing the knowledge ontology in a database. The method of claim 4, further comprising: providing a user with the right to add, delete, update, or give use of the knowledge ontology. The method of claim 1, wherein the user keyword network is established based on the information content of the user and the online information system. According to the method of claim 7, the step of constructing the user keyword network includes: recording the interactive browsing event of the user and the online information system into a log, and storing the browsing content. The method of claim 8, wherein the step of constructing the user keyword network further comprises: performing file exploration processing on the stored browsing content according to the record of the log of the user, and obtaining feedback vocabulary. The method of claim 9, wherein the step of constructing the user keyword network further comprises: establishing the user keyword network according to the feedback vocabulary and the knowledge ontology. The method of claim 9, wherein the step of constructing the user keyword network further comprises: for a keyword of the user keyword network, according to the browsing content of the keyword, Associating a plurality of attributes includes a time parameter and a spatial parameter. The method of claim 11, wherein the user keyword network comprises structured content of a plurality of nodes, each node corresponding to at least one keyword and a corresponding weight value. The method of claim 11, wherein the method further comprises: storing the user keyword network in a database. The method of claim 1, wherein the user keyword network comprises structured content of a plurality of nodes, each node corresponding to at least one keyword and a corresponding weight value. The method of claim 14, wherein the keyword of the user keyword network is associated with a plurality of attributes including a time parameter and a spatial parameter according to the browsing content from the keyword. The method of claim 1, wherein the method further comprises: providing a user with a new, delete, and update control function, a right for each node, and a control function and permission for opening to other users at different levels of use. . The method of claim 1, wherein the step of performing the user inference includes: generating, according to the service demand, information related to the service demand from the user keyword network, generating The content of the variation analysis; the association between the keywords of the content of the variation analysis is inferred to infer the relevance of the keywords to generate the user's intention to perceive the content. The method of claim 17, wherein the step of generating the content of the variation analysis comprises: intercepting data content related to the service demand in the user keyword network according to the service requirement, and merging in a tree form The algorithm performs a merge or a growth calculus to generate the variance analysis content. The method of claim 17, wherein the step of performing the inference includes: performing a user inference based on the inference method according to the content of the variation analysis, and generating a keyword relationship structure as the user. Intent to perceive content. The method of claim 17, wherein the method further comprises: storing the user's intention to perceive the content in a database. A computer readable recording medium storing a code, and when an electronic device or system loads the code and executes it, the method of any one of claims 1 to 20 can be achieved. An electronic device having a user-intentioned intention includes: a user interface unit for interactively presenting data; and a processing module for viewing content viewed by the user interface unit and at least one knowledge ontology; Constructing a user keyword network; wherein the processing module responds to a service requirement, and extracts data content related to the service requirement in the user keyword network via mutation analysis, and performs user intention inference to generate a user Intent to perceive content. The electronic device of claim 22, wherein the processing module establishes the user keyword network according to the knowledge body based on the information content that the user interacts with the online information system. The electronic device of claim 22, wherein the user keyword network comprises structured content of a plurality of nodes, each node corresponding to at least one keyword and a corresponding weight value. The electronic device according to claim 24, wherein the keyword of the user keyword network is browsed by the keyword, and the plurality of attributes are associated with a time parameter and a spatial parameter. A data computing system having a user-awareness, comprising: a communication unit for receiving content and events browsed by a user; and a processing unit for receiving content and events viewed by the user And at least one knowledge ontology, constructing a user keyword network; wherein the processing unit responds to a service requirement, and extracts, by using the mutation analysis, the data content related to the service requirement in the user keyword network, and performs user intention inference To generate user intent to perceive content. The data computing system of claim 26, wherein the processing unit establishes the user keyword network according to the knowledge body based on the information content that the user interacts with the online information system. The data computing system of claim 26, wherein the user keyword network comprises structured content of a plurality of nodes, each node corresponding to at least one keyword and a corresponding weight value. The data computing system of claim 26, wherein the keyword of the user keyword network is browsed by the keyword, and the plurality of attributes are associated with a time parameter and a spatial parameter.