Computer system and computer-implemented method for a guided query

The invention relates to a computer system and computer- implemented method for a query .

Up to now, professionals and experts in digitali zation like architects or developers , rely on their experience , their immediate social network, or utili ze social networks such as Yammer or Stackoverflow, or use other web resources such as internal websites or search engines to come up with answers . In many cases , the websites are either speciali zed ( e . g . Stackoverflow) or are generic such as search engines or social networks . In the case of the search engines , the information topic relevant to an expert is not organi zed suf ficiently or the relevant information may be missing . In most cases , search engines do not provide functions to speci fy context . Getting too many generic or irrelevant results is also a problem with search engines .

The topic-speci fic websites on the other hand, such as intranet pages , may have only a subset of the information required to solve a particular problem . An architect , for example , must currently manually go through many search results to build up an understanding of the solution and must reconstruct the implicit knowledge available across multiple information sources for the same problem each time anew . In the case of topic speci fic information sources , the architect must refer to sources manually to extract relevant information for their requirements .

Known recommendation systems , chatbots based or machine learning do not provide enough data in speci fic use cases , as the domain may be very speciali zed, without statistically irrelevant data availability . Each request may have its own unique context . Catalogues with filtering abilities provide too many options to choose from at once .

Accordingly, it is an obj ect to improve queries on topic speci fic use cases , in particular for digitali zation and software development . Furthermore , it is an obj ect to improve query systems .

This obj ect is achieved by the features of the independent claims . Advantageous embodiments emerge from the dependent claims .

According to an aspect of the invention, a computer- implemented method and computer system for a guided query comprising a graph database in combination with a context memory is proposed . According to a further aspect , a guidance algorithm is provided as well as a computer program product and a provision apparatus .

The invention provides according to the first aspect a computer system for a guided query comprising the components : frontend software component for providing guiding questions to a user, storage component comprising a graph data base containing guidance options with links , guidance software component comprising a guidance algorithm providing an option selection and retrieval of data associated with a selected option, wherein the guidance algorithm enables to query the graph data base and wherein the guidance algorithm enables creating, updating and querying a user speci fic context memory on a further storage component . The further storage component for the context memory is integrated in or accessible by the guidance system .

According to a further aspect , the guidance software component is stored and executed on a computer, wherin the computer can be a local computing system or web service . According to a preferred embodiment of the computer system, the graph data structure comprises a guidance graph, where guidance root nodes represent options at which a guidance session begins , wherein guidance session corresponds to the provided method .

According to another exemplary embodiment of the computer system, the guidance graph comprises at least one guidance root node , where the guidance starts .

In particular, the computer system comprises option nodes that can trigger di f ferent option-speci fic actions .

Preferably, the computer system according to an aspect of the invention, comprises directional connections between the guidance nodes representing the relations , which are used to resolve recommendations for a subset of chosen nodes .

The proposed guidance system provides a unique way of utili zing graphs to provide guidance for professionals . It not only overcomes the deficiencies of present knowledge graphs and decision trees , it rather takes these as a basis for research and explore semantic connections and adds a superstructure to it , that provides a stateful system that navigates the graph .

While a common knowledge graph traversal , e . g . browsing Wikipedia, would quickly increase the number of articles to read, taking into account the context of a query, the provided system narrows down the scope . The inventive guidance system provides traversal through multiple paths , with considering the context of the query type .

The invention provides according to a further aspect a computer-implemented method for a guidance session comprising the measures : receive a user' s request , select a guidance root node based on the user' s request by accessing a graph data base , wherein the graph data base provides guidance options with links , group in an iterative manner guidance op- tions by type and put them into a queue , dequeue each option group and check each option' s relevance in the context , provide the option to the user and provide a first guiding question, store user speci fic interactions history in the context memory and/or perform option-speci fic actions .

Such guidance session is representing a guided query respectively, at which end option-speci fic actions are performed, that can be information output or providing a link . In particular, the computer-implemented method further comprises the measure of providing links to knowledge sources , providing an answer to the user or start an application . Preferably, the measures are performed or kicked of f by the guidance algorithm .

The graph can have multiple entry points . In an exemplary embodiment of the method the guidance starts at any existing guidance root node , see description of Figure 2 in which a data model illustration is given . The multiple entry points allow flexibility of use of the guidance system .

The invention provides according to a further aspect a guidance algorithm performing the measures of a computer-implemented method according to any embodiment of the invention, performing the steps : start by selecting a guidance root node , leading to retrieve options connected to the selected root node , then group in an iterative manner retrieved options to option groups which is followed by the step queue the option groups followed by dequeue an option group while the queue is not empty, followed by relevance decision in the user speci fic context of the guidance session, in an iterative manner for each option of a dequeued option group, go to next option iteration in case of no relevance , then present option to the user in case of relevance , store option in the context memory i f chosen by the user and perform option-speci fic action, then retrieve options connected to the chosen option and with the retrieved options and go to next iteration of the group and queue step . In an advantageous embodiment of the guidance algorithm the guidance options are grouped by type .

The guidance algorithm does not traverse a unique path through the graph, but rather multiple paths at the same time . The proposed solution does not depend on machine learning techniques since there is not enough relevant internal knowledge-based data to train machine learning algorithms . Instead, the proposed guidance system stores the knowledge explicitly in the form of graph . The user speci fic information is generated in the form of context at each step of guidance session and is used to further traverse the graph . Concurrently, the Guidance System or the guidance session does not overwhelm the user with too many options but queues them in order to interrogate the user in a sequential fashion .

A further aspect of the present invention provides a computer program product comprising instructions which, when the program is executed by a computer, cause the computer to carry out the steps of the method disclosed by the present disclosure .

Furthermore , a provision apparatus for the aforementioned computer program product is claimed wherein the provision apparatus stores and/or provides the computer program product , e . g . saves and/or provides a program data and/or instruction block of said instructions .

The providing apparatus can be a data storage saving and/or providing the computer program product . Alternatively, the providing apparatus can be a computer system and/or a server system and/or a network and/or a cloud-based computer system and/or virtual computer system . The providing apparatus saves and/or provides the computer program product . Preferably, the complete computer program product is provided by said providing apparatus in the form of a download, for example , as a file or a data stream . Alternatively, the computer program product is provided by at least two partial downloads , for example , provided by a peer to peer network, each containing a piece of said computer program product . For example such a computer program product can be provided by a data storage , which is read and executed by a computer system . As a consequence the computer system is able to execute the disclosed method . Alternatively and/or additionally, the computer program configures a manufacturing device as above mentioned .

A data storage or computer-readable medium can be electronic, magnetic, optical , or a semiconductor system ( or apparatus or device ) . Examples of a computer-readable medium include , but are not limited to , a semiconductor or solid state memory, magnetic tape , a removable computer diskette , a random access memory (RAM) , a read-only memory (ROM) , a rigid magnetic disk, an optical disk, etc . Current examples of optical disks include compact disk-read only memory ( CD-ROM) , compact disk- read/write ( CD-R/W) and digital versatile disk ( DVD) .

Alternatively and/or additionally, the provision apparatus is a network service , a computer system, a server system, particularly a distributed computer system, a cloud-based computer system and/or a virtual computer system that stores and/or provides the computer program product preferably in the form of a data stream, for example . This provision is effected as a download in the form of a program data block and/or instruction data block, preferably as a file , particularly as a download file , or a data stream, particularly as a download data stream, of the complete computer program product , for example . This provision can alternatively be ef fected as a partial download that consists of multiple parts and is downloaded particularly via a peer-to-peer network or provided as a data stream, for example . Such a computer program product is read in, for example using the provision apparatus in the form of the data storage medium, in a system and exe- cutes the program instructions, so that the method according to the invention is executed on a computer, or configures the creation device such that it creates cited system and/or execution unit according to the invention.

Preferably, the presented embodiments are implemented by a processor and/or a memory device unless otherwise noted.

In detail, to implement and/or execute the invention, the inventive method, components, devices etc. comprise at least one processor and/or at least one memory device unless otherwise noted. Additionally, the inventive method, components, devices etc. comprise, for example, other features known by a skilled person. For example, these features can be an input device, like a computer mouse, or a display device, like a TFT-display .

In another preferred variant of the invention, a processor is specifically configured to execute the computer readable instructions such that said processor is preferably configured to perform functions which implement a step/all steps of the inventive method.

Definitions :

As used herein, the term "graph data structure" is a database that uses graph structures for semantic queries with nodes, edges, and properties to represent and store data, database (GDB)

Unless specifically stated otherwise, as apparent from the following discussions, it is appreciated that throughout the specification discussions utilizing terms such as "processing", "computing", "computer-based", "calculating", "determining", "generating", "configuring" or the like, refer to the action and/or processes of a computer that manipulates and/or transforms data into other data, said data represented as physical, e.g. such as electronic, quantities. The term "computer" should be expansively construed to cover any kind of electronic device with data processing capabilities, including, by way of non-limiting example, personal computers, servers, handheld computer systems, Pocket PC devices, Cellular communication device and other communication devices with computing capabilities, processors and other electronic computing devices.

As used herein, the term "processor" refers, for example, to controllers, microcontrollers (e.g. digital signal processor (DSP) possibly in combination with memory and storage units memory unit storing computer-readable instructions, application specific integrated circuit "ASIC", etc.) , processor cores, central processing units, integrated cir- cuits/processing circuitry (e.g. application specific integrated circuit "ASIC", field programmable gate arrays "FPGA" etc.) or digital signal processors. Furthermore, the term "processor" can, for example, refer to virtual processors, virtual CPUs, soft processors or soft CPUs. Moreover, said processor can, for example, be configured to execute computer readable instructions such that said processor is preferably configured to perform functions which implement the invention .

As used herein, the term "component" refers, for example, to a processor and/or a memory unit storing computer-readable instructions. For example, the processor is specifically configured to execute the computer readable instructions such that said processor is preferably configured to perform functions which implement the invention, such as a step of the inventive method. Furthermore, the term "component" can, for example, refer to means (e.g., a processor) which are configured to implement/execute functions/steps of the invention.

As used herein, the term "subtree", "tree" or the like, refer, for example, to data structures storing information about the technical system and/or components of the technical system . Preferably, a subtree is a branch of a ( component fault ) tree or a partial ( component fault ) tree , defined, for example , by a selected node in the tree as top-level node of the subtree .

In the following, possible embodiments of the di f ferent aspects of the present disclosure are described in more detail with reference to the enclosed figures .

The accompanying drawings are intended to provide a better understanding of the embodiments . They show embodiments and serve in conj unction with the description to explain principles and concepts of the disclosed subj ect matter . Other embodiments and many of the cited advantages will emerge in respect of the drawings . The elements of the drawings are not necessarily shown in scale with one another . Identical reference characters here designate identical components or components of comparable ef fect .

The properties , features and advantages of this invention that are described above and the manner in which they are achieved will become clearer and more distinctly comprehensible in conj unction with the description of the exemplary embodiments that follows , these being explained in more detail in conj unction with the figures 1 to 3 , in a schematic depiction :

Fig . 1 shows a diagram of an exemplary embodiment of a guidance computer system according to the invention,

Fig . 2 shows a flowchart representing an exemplary advantageous embodiment of the guidance system data model , Fig . 3 shows a flowchart representing an exemplary workflow of an embodiment of the guidance algorithm .

As shown by the diagram in Figure 1 , an exemplary embodiment of the guidance system provides the following features :

The proposed Guidance System GUID incrementally guides a user USE by providing relevant guiding questions and suggestions . The actual knowledge may reside in other systems , for example databases EXT , internal repositories , social networking websites SERV etc but the Guidance System GUID contains explicit links 1 towards the knowledge KNOW, like solutions , templates , articles and/or actions to take . The knowledge in the Guidance System GUID is stored in the form of a graph, compare Figure 2 , where the nodes provide context or content for the guidance .

Figure 2 shows a flowchart representing an exemplary workflow of an advantageous embodiment of the guidance system GUID . Nodes are for example root nodes , rl , r2 , rn, where the query starts . The context available in the nodes is used to traverse the graph to guide the user USE to context relevant information, res l-resn .

During a guidance session, the graph is incrementally traversed, starting from a dedicated starting node rl , r2 or rn, a so-called guidance root . Nodes can be added to the guidance context via any form of interactive feedback with the user USE .

The guidance system GUID is extendable via the underlying data . Each additional new data source information res 1 , ... resn can be added to the Guidance System graph GUID as nodes . In case the categories of the additional data do not require a modi fication of the user interface , the guidance software doesn' t even need to be updated .

I f several available options ol , ...on exceed a reasonable number, then moderation, filtering based on additional algorithms , pagination, postponing the options ol , ... on for later, can be applied to keep the user experience lean .

A dedicated user interface is for publishing the knowledge arti facts and their relations .

Basically, a guidance session follows a procedure , presented in Figures 2 and 3 : given a guidance root rl , r2 , select all the options otl-ot4 pointing towards that root rl , r2 and group og2 / og3 the options via their tag otl , ... ot4 . This tag could be a deployment option cd/pd or an action sea/bui . Then queue qu the option groups og2 /og3 , compare Figure 3 .

While queue not empty dec-n, the next step is to dequeue deq the option group . Only the options ol , ... on are presented to the user USE , which are relevant given the context of the guidance interaction, e . g . which are connected, that means directed towards the selected ones to the already selected options or to all options , i f none option is selected yet . The selected option than is added to the selected options list .

The option nodes and connections can have properties associat ed with them that can be used to trigger actions , such as sea rching for API s , presenting a special widget or showing links 1 to resources res l-resn . This association with custom actions is very powerful and enables various scenarios .

An architect for example might use a Guidance System GUID as exemplary presented in Figure 2 , accessing the system via a webpage rl , to query for information for his use case . This begins a guidance session with the architect . The architect can now select among the multiple options ol , ... on like "Build an App" bui or "Search for keywords" sea . Based on option selected at each guidance session interaction, the next possible options are suggested . The example above is one such data model used in a guidance system .

The interaction between the architect USE and the Guidance System GUID can look like the following exemplary chat in Tabs 1 and 2 . The last two chat bubbles of Tab . l provide the two available options for the user USE of the system GUID . Tab. 1 shows a first chat box with an exemplary chat:

Chatbot : User :

Hello, Welcome to the Guidance

System

Please enter your query, and we'll see how we can help. building sensors

You've requested 'building sensors' . What would you like to do? build an app for your concern just search for the keywords

The next sample guidance session in Tab.2 shows a complete interaction, which results in suggestions to browse to documentation resources resl, ...resn. One of the links 1, e.g. to Mendix app on Mindsphere is shown only because earlier in the discussion, the question about the deployment in the Cloud was answered with Mindsphere, and not Azure, for exam- pie. This is what demonstrates the aforementioned context of the guidance session, having an influence on the proposed information sources resl, ... resn. Tab. 2 shows a second chat box with an exemplary chat:

Chatbot: User:

Hello, Welcome to the Guidance System Please enter your query, and we'll see how we can help. building Sensors You've requested 'building Sensors' . What would you like to do? build an app for your concern

You've requested 'build an app for your concern' . What would you like to do? no programmers available You've requested 'no programmers available' . What would you like to do? deploy in the Cloud You've requested 'deploy in the Cloud' .

What would you like to do?

Mindsphere Cloud You've requested 'Mindsphere Cloud' . What would you like to do? build and deploy a Mendix app

Visit the Mendix Homepage for more Information

Learn how to deploy a Mendix app on

Mindsphere

You've requested 'build and deploy a Mendix app' . What would you like to do? use the Data Layer as the data source

Figure 3 shows a flowchart representing an exemplary workflow of an advantageous embodiment of the algorithm.

An exemplary algorithm in pseudocode could read: Start Guidance Session

Choose a Guidance Root node

Query Options connected to the root node

Group selected Options by type and put them into the queue WHILE Queue NOT Empty

DO Dequeue an Option Group ( an algorithm may be used to choose a group to dequeue )

FOR each Option in the Option Group

I F Option relevant in the Context ( an algorithm may be used to determine relevancy) THEN Present the Option to the user ENDI F

ENDFOR

User chooses an Option out of the presented ones Store the chosen Option in the Context Store Perform Option-speci fic actions

Query Options connected to the Chosen Option

Group selected Options by type and put them into the queue

ENDWHILE

This describes an algorithm based on a guidance graph, where nodes represent options ol , ... on to be guided through or towards . The directional connections between the guidance nodes represent the relations , such is IS_OPTION, which are used to resolve recommendations for a subset of chosen nodes . The guidance starts from a special node , which is a guidance root rl , r2 , ... rn . There can be multiple guidance roots rl , r2 , ... rn in the graph, for di f ferent use cases .

In summary, a guidance system GUID is proposed that can be used by professionals , for example architects , developers , managers , etc . to query for their digitali zation use cases . The system can guide the user USE by directing them to relevant information res l , ...resn based on user speci fic context as well as query specific context. This query specific context refers to the previous interactions in the guidance session which further improves providing relevant information. The guidance system GUID has access to a graph, e.g. a graph database and a context memory. The graph database especially stores information about other knowledge sources KNOW. In the context memory, the context related information to a specific user querying the guidance system is stored.

References :

USE user

PC personal computer req request

GUID guidance system, e.g. including storage of graph database and including context memory

STOR storage component for storage of graph database and context memory, can be in eluded in the guidance system

1 link

KNOW storage for knowledge source, e.g. Wiki, Code, Catalogue SERV server, connection to communication network or search service

API external API rl, r2, ... rn guidance root 1, 2, n: examples - web pages, starter project/ wizard otl, ot2, ... otn option type ol, o2, ...on option og2/og3 ...ogn option group sea search for keywords bui build an app 1c low code pr programmers available PM Programming model cd cloud deployment pd on premise deployment DM Deployment model

CPI, CP2, ...CPn cloud platform 1, 2, ... n: Examples: Siemens Mindsphere, Microsoft azure; Cloud product Mendix is deployed on siemens mindsphere or cloud product data layer is deployed on Microsoft azure . prod 1, 2, n cloud product 1, 2, examples: Mendix, data layer, datafinity resl, res2, ... n external resource 1, 2, 3, 4, examples: mendix homepage, Mendix on Mindsphere, Data Layer Starter Project, Datafinity Homepage st start sei select / select a guidance root node ret, ret2 Retrieve grp, qu group guidance options by type and put them into the queue dec decide y yes n no dq dequeue / dequeue an option group next o go to next option iteration rel relevance check, whether option is relevant in the context show show to the user store storing in the context memory pert perform / perform option-specific actions end end