DE102013109295A1 - Method for evaluating data and analysis system - Google Patents

Abstract

The invention relates to a method for evaluating data and an analysis system for carrying out the method, wherein the data are stored electronically. At least two generic terms (2) are subdivided into non-overlapping categories (3, 4), whereby the data per generic term (2) is assigned to a category (3, 4). On a asked query, which includes a target size and at least one category (3, 4), a target value (6) is delivered as a search response (5) To reduce unnecessary system load, for example by repeated inputs of the same or similar searches, automatically additional search responses (7, 8), which provide target values for modified search queries, in which the target size and / or one of the categories (3, 4) contained therein is varied compared to the given search query, of the additionally generated search responses (7, 8) only the search responses are delivered whose target values lie in an edge region of the set of the determined target values, wherein the underlying modified search queries are stored for these additionally generated and supplied search responses (7, 8)

Description

The invention relates to a method for evaluating and displaying data according to the preamble of claim 1 and to an analysis system according to claim 10.

Such methods or analysis systems are used in particular for the evaluation of large amounts of data. The method is often implemented in software known as BI tools, which provides access for appropriately trained users.

The data must first be categorized in order to be used for evaluation, for example in generic terms and categories. This achieves a structuring of the data on the basis of hierarchies and enables a simplified evaluation.

For the evaluation of the data, a search request is made by the user, wherein the search request includes a target size and the specification of the categories in which to search. The target size indicates what is being searched for. For example, these are key indicators, so-called KPIs.

The search response is then a target value, that is usually a specific numerical value or a percentage indication of the target size in the or the corresponding categories.

For a more specific specification and limitation of the search response, a combination of categories can also be used in the search query. With multidimensional structures, relatively complex search queries result. Due to the high complexity, the problem arises of making mistakes in the creation of search queries, which can thus lead to a search response that does not fit into the context and thus to a misinterpretation of the data.

Another problem is that it is only possible to search for concrete, ie known, target variables. An inaccurate knowledge of the target values can therefore lead to the result that the correct search response can not be determined or at least several attempts are required until the correct search response is obtained. On the one hand, this is time-consuming and, by making many unnecessary search queries, results in an unnecessary load on the system, for example a server, as is generally used to integrate a database in which the data is stored.

In order to avoid creating an incorrect search query, for example by selecting an incorrect category, and thus not detecting essential insights from the data, it is often necessary to approach the desired search query by asking more queries of increasing complexity. However, this in turn leads to a relatively high system load, for example due to the largely overlapping search queries, which increase a need for computing power.

The invention is based on the object to avoid the disadvantages of the prior art and in particular to provide a method or an analysis system that avoids an unnecessary system load. Furthermore, the risk of misinformation for the user is to be reduced by inaccurately asked search queries and the user may be made aware of additional findings that can be obtained from the data, which are related to his search request. The use of the method or the operation of the analysis system should thereby also be possible by less trained users and the effort to gain knowledge can be reduced.

This object is achieved by a method according to claim 1 and by an analysis system according to claim 10. Main features of the invention are given in the characterizing part of claim 1 and of claim 10. Embodiments are the subject of claims 2 to 9.

In a method for evaluating data stored electronically, wherein at least two generic terms are subdivided into non-overlapping categories, and the data is (exactly) assigned to a category per generic term, wherein a search query that includes a target size and at least one category, is provided as a search response, a target value is provided according to the invention to generate additional search responses that provide target values for modified searches in which compared to the search query, the target size and / or one of the categories contained therein is varied, of the additional search queries generated in particular only the search responses are delivered whose target values lie in an edge region of the set of the determined target values, the underlying modified search queries being stored for these additionally generated and supplied search responses.

With the aid of the method according to the invention, therefore, only a single search request has to be provided by the user, but not only the associated search response is delivered to him, but also related relatives located in the same context Search answers. The number of search responses is thereby limited to a manageable amount so that only the target values lying in the border area are displayed. It is assumed that these target values lying in the peripheral area have a high relevance. Overall, the space available for example on a display is optimally utilized. In addition, for example, the user can be alerted to significant changes in the data that often indicate problems. The method according to the invention thus not only leads to a reduction of the search queries to be provided and thus to a reduction in the system load, but also offers optimized information transfer and thus increased security. In this case, by storing the generated search responses with the associated search queries, a multiple calculation can be avoided by making it possible to resort directly to the already stored search response when such a search request is entered again.

Since not only the additionally generated search responses are specified, but also the associated search queries, a kind of backward search can be performed. The user not only receives the information about a certain target value, but also in which categories, ie for which data, this is correct. In this way, the user can, if appropriate, specifically search for further outcomes in these categories.

A more precise narrowing of the search response can be obtained by the search request comprising a combination of categories of different generic terms. Thus, a multi-dimensional search is possible, which can lead to correspondingly accurate results.

By specifying the selectable targets and / or categories, user-friendliness can be increased and less professional users can be offered meaningful search queries. Search queries that, for example, do not lead to useful search responses due to a lack of suitable data and thus only represent an unnecessary burden on the system, are thus prevented.

A real-time presentation of the search responses can be achieved by generating and storing the search responses before placing a search query by forming the possible combinations of target values and categories. The otherwise inevitable due to the required computing time waiting after placing the search can be minimized. In this case, the pre-calculation of the search responses, for example, even in times where no search queries are currently being made. This results in a higher utilization of the system and a reduction of power peaks.

In a preferred development, only combinations of target variables and categories are provided for which a search response is available. Where appropriate, proposals for possible combinations may also be made. For example, an auto-completion when entering the search request can be provided. This increases the user-friendliness. In addition, the burden on the system is kept low because search queries for which search queries can not be generated need not be processed at all.

The target variables and / or categories used in the modified search queries are preferably varied on the basis of stored relationships of the target variables with one another and with the categories. For example, the relationships are also stored in a database and determined from empirical values. In this way, it can be ensured that matching target values or search responses are output only in the same context, and thus no exposure is caused by undesired search responses.

Preferably, target values are assigned to the target size, whereby the target values and / or the related target values are classified on the basis of the target value rules. For example, it may be provided to colorize the target value if it is within a certain interval. The visual recording of the information is thus simplified. At the same time a rating can be displayed, for example by a red color for a rather negative target value and a green color for a more positive target value.

Preferably, the generated queries are filtered to belong to individual headings and / or categories. This makes it relatively easy to make further search queries based on the search responses already received by the generated search queries and thus, without having to make multiple search queries multiple times, to refine the search.

For a simple and reliable evaluation of the data, it is advantageous that the data are assigned to the generic terms and categories according to the MECE principle. There is a clear, non-overlapping categorization of all data in tree structures. Several trees or tree structures can be created starting from a generic term in each case, whereby the data in each tree is assigned to individual categories without any overlap. Each tree forms a dimension in which the entirety of the data can be considered. Accordingly, several trees allow a multi-dimensional view and thus a refined analysis of the data. The methodical and systematic use of MECE-compliant tree structures makes even complex data structures easy to understand.

The object stated in the introduction is achieved by an analysis system for carrying out the method according to the invention in that it has a memory for holding the data, a computation component for processing the search queries and for developing the search responses, a display for displaying the search queries and an input interface for entering the search queries having.

The memory can have a database structure and be connected, for example, via an online access to the computing component. The display is advantageously carried out via a screen, which may optionally be designed as a touch screen and thus has an integrated input interface. This can also be designed independently, for example, as a keyboard.

For the analysis system, the features and advantages mentioned in connection with the assessment of the method apply analogously.

Further features, details and advantages of the invention will become apparent from the wording of the claims and from the following description of exemplary embodiments with reference to the drawings. Show it:

1 a basic structure of a tree structure,

2 a schematic representation of a search response with additional generated search and

3 an analysis system.

In 1 is the schematic structure of a tree structure 1 shown. A generic term 2 is clearly in categories 3 . 4 divided, in turn, into categories 3.1 . 3.2 . 3.3 respectively. 4.1 . 4.2 are divided. This can then be followed by other subordinate categories. Data can thus be assigned to a category without any overlapping, with increasing category levels resulting in an ever finer subdivision. Parent categories always include all data of the lower categories.

With the help of several tree structures, which each represent a tree and are applied to the same database, a multi-dimensional viewing and evaluation of the data is possible. The assignment of the data is carried out according to the MECE principle.

For example, a tree structure is constructed starting from the generic term location. The categories 3 . 4 could then be, for example, continents that make up other categories 3.1 . 3.2 . 3.4 . 4.1 . 4.2 such as countries. A further subdivision by additional categories could be states / regions, districts, municipalities, cities, neighborhoods, streets, etc., so that a very fine-grained categorization of the data is possible.

Another tree structure for viewing the same data from another dimension could, for example, include the term thing as a generic term, which is then e.g. into the categories of vehicles, land vehicles, motor vehicles, passenger cars, vehicle class, ie upper class / middle class, etc. is divided. These data are then also assigned without overlapping.

When evaluating the data by means of a corresponding search query, as it were, the tree structures or categories of different tree structures are superimposed, whereby an intersection of the data is obtained. A target value belonging to a target variable is then determined for this intersection.

For a complete search usually includes the specification of the target size, so what should be searched. For example, search for the number of new registrations in a specific city and for a particular vehicle class. The target is the new registrations.

The target value output in a search response is then the actual number as a numeric number.

Often, a single search response or a single target value is not meaningful enough. According to the prior art approach, therefore, additional search queries have to be provided by a user and the resulting search responses must be placed in a common context. This is where the invention starts.

Additional search queries are automatically generated by targeting the entered search query and evaluating it. For example, one will always be one with the Search query entered (target size, categories) exchanged and evaluated.

So that an unreasonable number of additional search responses is not presented to the user, it is intended to display only those which lie in an edge area of the determined target values. In particular, the underlying search parameters, that is the modified search query, are specified in the search response for these target values.

Staying with the above example, for example, it could be displayed as an additionally generated search answer, in which cities the most or the fewest approvals were made in the same vehicle class. For example, another search response might consider historical data and indicate how much the number of registrations has changed.

A possible display of the search response and the additionally generated search responses is in 2 shown schematically. The search response determined on the specific search query 5 comes with the associated target value 6 specified. However, the user receives additional information through additionally generated search responses 7 . 8th , which are graphically represented by stored target value rules. The various search responses, ie the information obtained, can be in the form of a score card 9 be summarized.

For the additionally generated search response 7 For example, it is the display of a trend. As a target value rule is then deposited, with increasing registration numbers to indicate an upward arrow. The further search answer 8th is designed in the form of a traffic light and displays, for example, a green color if the number of registrations is higher than average in other cities, while a red color is displayed if the number of registrations is less than the average number of registrations in other cities.

In principle, the target value rules can be designed, inter alia, in a competitive, plan-oriented or time-oriented manner and thus serve to obtain information quickly by specifying, for example, the graphical representation.

Depending on the available data and the respective application area, different, context-related additional search responses can be generated according to this principle and displayed in addition to the requested search response. The number of search queries to be entered by the user is thus significantly reduced. In addition, the user receives the desired information in a much shorter time.

In 3 is an analysis system 10 for carrying out the method according to the invention shown schematically that a memory 11 having a database for storing the data. The memory 11 is for example via an online connection with a computing component 12 connected to the one ad 13 connected. An input interface is used to enter the search requests and, if necessary, further commands 14 ,

The invention is not limited to one of the above-described embodiments, but can be modified in many ways. For example, it may additionally be provided to predefine search queries contextually. Also, especially when search queries are precalculated for all possible combinations of target variables and categories, provision may be made for permitting search queries directed specifically to target values or even open search queries.

By means of the method according to the invention or the analysis system according to the invention, a BI tool is obtained in which the number of search responses to be entered is reduced and thus a system load is reduced. It is also possible for less experienced users to receive appropriate search responses and to evaluate the data. The users do not have to have any in-depth knowledge of statistics in order to interpret the search responses. In particular, not only an analysis is carried out in relation to the query, but also an analysis of the environment. In addition, the optimum combination of categories for a target value can be found by displaying the search queries underlying the additionally generated search responses. Optionally, the optimal combinations for different views can also be found.

All of the claims, the description and the drawings resulting features and advantages, including design details, spatial arrangements and method steps may be essential to the invention both in itself and in various combinations.

LIST OF REFERENCE NUMBERS

1

Threaded

2

preamble

3

category

4

category

5

search response

6

target value

7

Additional search response

8th

Additional search response

9

Scorecard

10

 analysis system

11

 Storage

12

 computational component

13

 display

14

 Input interface

Claims (10)

Method for evaluating data stored electronically, wherein at least two generic terms ( 2 ) in non-overlapping categories ( 3 . 4 ) and the data per generic term ( 2 ) of a category ( 3 . 4 ), based on a search request, a target size and at least one category ( 3 . 4 ), as search response ( 5 ) a target value ( 6 ), characterized in that additional search responses ( 7 . 8th ) that provide target values for modified queries that have the target size and / or category (s) compared to the given query ( 3 . 4 ), whereby of the additionally generated search responses ( 7 . 8th ), in particular, only those search responses are delivered whose target values lie in an edge region of the set of the determined target values, whereby these additionally generated and supplied search responses ( 7 . 8th ) the underlying modified search queries are deposited. Method according to claim 1, characterized in that the search query is a combination of categories ( 3 . 4 ) different generic terms ( 2 ). Method according to claim 1 or 2, characterized in that the selectable target variables and / or categories ( 3 . 4 ). Method according to one of the preceding claims, characterized in that the search responses before placing a search query by forming the possible combinations of target variables and categories ( 3 . 4 ) are generated and stored. Method according to one of the preceding claims, characterized in that only combinations of target variables and categories ( 3 . 4 ) for which a search response is available. Method according to one of the preceding claims, characterized in that the target variables and / or categories used in the modified search queries ( 3 . 4 ) on the basis of underlying relationships between the target variables and the categories ( 3 . 4 ) can be varied with each other. Method according to one of the preceding claims, characterized in that target values are assigned to the target variable, wherein a classification of the target values and / or the related target values takes place on the basis of the target value rules. Method according to one of the preceding claims, characterized in that the additionally generated search queries for belonging to individual generic terms and / or categories are filtered. Method according to one of the preceding claims, characterized in that the data are assigned to the generic terms and categories according to the MECE principle. Analysis system ( 10 ) for carrying out the method according to one of the preceding claims, characterized in that it has a memory ( 11 ) for receiving the data, a computational component ( 12 ) to process the search queries and to create the search responses, an ad ( 13 ) for displaying the search responses and an input interface ( 14 ) for entering the search requests.

Images