DE10024368A1 - Locating selection of information products involves accessing information product database containing data about information products with at least location, category information per product - Google Patents

Abstract

The method involves receiving a request with location and user identification information, accessing stored content information concerning the identified user, accessing an information product database (123) containing data about information products with at least location and category information for each product, assessing several products for a match to the user's wishes and selecting at least one product accordingly. Independent claims are also included for the following: a computer, especially an Internet server, and a mobile user device.

Description

The invention relates to the technical field, an automatic pre-selection within a wide range of information. Particularly suitable the invention for applications in which so many information offers Are available that a search or selection by a user is difficult is. This can e.g. B. be the case when information offers in one order comprehensive network such as the Internet. The stake is special the invention is intended for applications in which the user is the actual want to make a selection via a mobile user device that only has limited input and display options.

The Internet consists of a large number of stationary computers and computers networks that are connected to each other by communication links. The connected computers exchange information using various services from, for example, email, and WWW. The WWW service allows a server Computer system (web server, web site), graphic information pages on one remote user computer system (user device). Remove that te user device then represents these information pages (web pages).

Any resource, i.e. H. Computer or website in the WWW is through a URL (Uniform Resource Locator) clearly identified. To a specific web page the user device transmits the URL for the desired page in a request (e.g. via the hyper-text transfer protocol (HTTP)).

The request is then forwarded to the server computer that requested the Website offers. When this web server receives the request, it sends the requested website to the user device. If the user device the web page, it usually displays it using a browser uses. A browser is a special application program on the user device to request and display information pages.  

It is using wireless access protocols such as WAP (Wireless Access Protocol) possible, also from mobile user devices, e.g. B. from PDAs, cell phones, note books or organizers who use the information services offered on the Internet.

At the moment, web pages are usually created using the page description language HTML (Hypertext Markup Language) defined. HTML contains a stan standardized set of commands that can be used to determine how a WeV page is shown. For mobile devices that use WAP for transmission zen, there is another standard (WML), with a reduced instruction set and less graphic display options.

For the mobile user stand in contrast to users who use the Internet from from a permanently installed workstation, but information about the immediate surroundings in the foreground ("What is there here?"). For mobile users devices (e.g. WAP / UMTS cell phones) are "location dependent services" (location dependent services) that allow the user with few 'Clicks' or inputs from a specific provider in their vicinity (e.g. an open one Pharmacy).

The currently popular search engines for the Internet, however, work with a mere conversion from HTTP / HTML to WAP / WML and specify one query all available answers. This is usually a very large list, which is difficult on the small displays of today's mobile user devices is to be displayed.

The invention accordingly has the object of solving the problems of the prior art to avoid at least in part and to offer a way by the user select the information you are looking for as reliably as possible. In particular, the limited input and output options should mobile user devices are taken into account.

According to the invention, this object is achieved by a method with the features of Claim 1, a computer with the features of claim 14 and a mobile User device with the features of claim 15 solved. The dependent types sayings relate to preferred embodiments of the invention.  

The invention is based on the basic idea of preselection based on least of location and user data (and in preferred embodiments additionally based on the time). The likely sought after should Information largely autonomous, in individual cases with just one entry action on the user device (e.g. pressing a key on the mobile phone) and are offered on the user device.

The invention drastically reduces not only the search effort, but it will also be a convenient search on a restricted (small ad, only few buttons etc.) enables user device. In preferred configurations the invention further enables an optimal representation of the offers and a simplified entry of search criteria.

The order of the process steps in the claims is not intended to be enumerated Restriction of the protection area mean. Rather, the steps of inventive method in a different order or in whole or partially parallel or interlocked.

In preferred embodiments, a computer or a computer system is provided as input parameters the current time, the position of the user and a user ID. The computer preselects the data from a database information that the user is likely to find (e.g. offers) and provides these are available on a user device before the user starts his search has entered the term. The preselected offers represent those of the user most likely to be the information you are looking for The choice made by the user flows back into the computer and thus influences the future preselections.

In advantageous developments of the invention, the location of the one a lot of contiguous areas or an area determines whose offers are considered in the further pre-selection process the. The amount of contiguous areas can in some areas events are increased until a minimum number of offers is received will.  

In preferred embodiments of the invention, the offers are categories assigned, and the categories form an acyclic directed graph, where a category points to the categories it contains.

In addition, embodiments are provided in which the preselection of loading keywords entered by the user are taken into account. The keywords can assign offer categories or offers in some configurations net, with the keywords category or offer designations fully or partially. Alternatively or additionally, can also by an evaluation of a log of the previous user selection behavior determine to what extent after entering a certain keyword Past users preferred certain offers or categories have chosen. It can then be the categories chosen with great frequency and offers are assigned to the keyword. It can also be provided be that only those offers are then given in the selected set remain that were either assigned directly to the keyword or that of Belong to the category assigned to the keyword.

If in some embodiments the time of the search query is in the previous selection, this point in time can be determined using the computer's clock become. The search query also contains a position as an input parameter information, it is provided in preferred embodiments, the by the Computer clock to correct the specified time using a time zone table ren that this is roughly the time of the request on the by the positions specified location (usually the location of the user device).

In preferred configurations, each offer is one or more times assigned sections that preferably correspond to the opening hours of the offer chen.

In order to evaluate or to exclude offers, preferential exclusions are used design rules that are user or user group specific and Anga ben about offer categories, offers and optional information about relevant Periods and areas included. For example, it can be provided that only the offers remain in the preselected offer quantity that cannot be excluded by a rule. Additionally or alternatively you can  in some embodiments, the offers are evaluated in the offer quantity by determining a number for each offer from the data of a rule, which is multiplied by a weighting factor and the valuation of the offer represents botes or is used to change the previous rating.

It is preferably provided that the data obtained from the protocol data are combined compiled statistics are used, the respective statistics containing data, how often or how likely an offer or offer category was chosen depending on the user and / or one used User profile (user group) and / or from one place and / or from one Territory and / or a day, week or season, and / or depending combination of the factors mentioned here.

In some embodiments of the invention, the offers are in the offer Quantity assessed by stating frequency or probability of the offer and / or a category to which the offer belongs and multiplied by a statistic-specific weighting factor. The way value obtained can then z. B. represent the evaluation of the offer and / or used to change the previous rating.

As a further development of the invention, it can be provided that the evaluation interrupted by rules or statistical data when a time limit over steps have been taken or the offers have been assessed with sufficient differentiation.

In preferred embodiments, the offer quantity after the evaluation changed so that only those offers remain that have a minimum rating exceed. It can also be provided that the number of offers in the Limit the quantity to a maximum number and the quantity after the evaluation to be changed in such a way that only the highest-rated offers remain included.

Provision is preferably made for preparing the offers for presentation evaluated amount of offers so one of linked information pages transform existing tree that the highest rated deals on appear on the first page.  

In some configurations, if there are only a few offers in the pre-selection a category can be found, this category completely omitted.

With regard to further preferred refinements of the method, is again expressly referred to the dependent claims, the content of which is hereby incorporated by reference full scope is also included in the present description.

The computer according to the invention and the user device according to the invention are preferably further developed with features that the just described and / or correspond to the features mentioned in the dependent procedural claims. The mobile user device can be used to enter parameters (e.g. location, key words, user identification,. . .) be set up.

Further features, advantages and objects of the invention are evident from the following detailed description of several embodiments and from the scheme table drawings. Show it:

Fig. 1 is a schematic illustration of an exemplary system configuration,

Fig. 2 shows an overview like flowchart of an embodiment of the inventive method,

Fig. 3 illustrates an exemplary table of the offered information,

Fig. 4 is a listing category graph

Fig. 5 shows an exemplary table for geo-information data,

Fig. 6 is a representation of variable regions

Fig. 7 is a detailed flow chart of the selection process,

Fig. 8 is a flowchart of the static selection by exclusion,

Fig. 9 is a flowchart of the selection based on the position information,

FIG. 10 is a flowchart of selection based on category,

Fig. 11 is a flowchart of the selection based on the sampling timing,

Fig. 12 is an exemplary illustration of a selection log file,

Fig. 13 is an exemplary illustration of user-specific statistical data,

Fig. 14 is an exemplary representation of user-specified rules,

Fig. 15 is an exemplary illustration of a result according to the evaluation,

Fig. 16 is a flowchart of the procedure for the preparation for presentation,

Fig. 17 is an exemplary representation of the result of the presentation preparation, and

Fig. 18 is a plan view of the guide, for example a keyboard of a user equipment in a corner.

Fig. 1 gives an overview of the system in the exemplary embodiment described here. The system contains one or more user devices 11 which are connected to a computer 12 via a network 13 , 14 , 15 . The network generally has a radio network 14 and a fixed network 15 . The user device establishes a connection to a transmission mast 13 via the radio network 14 . A second network connection, namely the landline 15 , leads from the transmission tower 13 to the computer 12 , possibly via several stations and intermediate devices, which are not shown in FIG. 1. As a result, the user device 11 is able to fetch data from the computer 12 and, conversely, the computer 12 can send data to the user device 11 .

The user device 11 has a browser 112 , which receives the incoming data via the radio network 14 and sends outgoing data to the radio network 14 . The data are shown in a display 111 of the user device 11 . Via an input component 113 of the user device 11 , e.g. For example, a keyboard, data can be entered, which the browser 112 then processes.

The computer 12 has a user database 112 , which contains data about all known users and their user profiles. An information offer database 123 contains data about offers. B. Shops with their addresses and opening times as well as offer profiles. A geo, place and street database 124 contain data about places and streets as well as about the geographical assignment of the streets among themselves z. B. to areas. This database also contains location profiles. Statistics and rules are stored in a statistics database 127 or in a rules database 126 . The computer 12 also has a clock 125 , which indicates the time on site.

A selection component 121 in the computer 12 receives data directly or indirectly from the user device 11 via the network 13 , 14 , 15 . This data can include information about the location, user and user device. Some of the data were entered by the user via the input component 113 or come from the browser 112 . The selection component 121 uses this data coming from the user equipment 11, and the profiles and data in the databases 122, 123, 124, compiling a selection of information from the information service database 123 for each user apparatus 11 and this then over the network 13, 14 To send 15 to this user device 11 . In the present exemplary embodiment, the time of the request based on the clock 125 is taken into account, for. B. if the information offered is a reference to a shop, for its opening times. If the location is known, the time at the user's device is also taken into account by adding a time difference. The browser 112 then displays the selection of information generated and sent by the computer 12 after receipt in the display 111 .

The invention thus enables the user of the user device 11 to use the display 111 to present an optimal selection of information from the information offer database 123 with regard to his or her current situation, ie time, personal preferences and his location. This selection is made by the selection component 121 .

The user selects one of the offers via the input component 113 . This user behavior is sent from the user device 11 to the computer 12 and processed there by a system process (shown in FIG. 2 with reference symbol 23 ).

As user device 11 come primarily mobile devices, e.g. B. mobile phones in question, but can also be used as user devices, laptops, notebooks, PCs, electronic organizers, navigation systems and palmtops, but it is also possible to use stationary computers, e.g. B. serve as a bank statement printer or ATM.

As network technology, GSM, GPRS and UMTS or future mobile Transmission standards are used. Networks are also possible in which only a fixed network or only a wireless network can be used (e.g. a BlueTooth network).

The invention is not limited to the WML data format, but can also without limitation for solutions based on HTML, XML or similar data formats based.

The computer 12 is preferably implemented by a powerful Internet server. However, a simple PC or laptop can also take on this role. With sufficient mobile computing capacity, a solution is also conceivable in which user device 11 and computer 12 are implemented in a single device. Although only one computer 12 is described here, the components of the computer 12 can also be distributed on different interconnected computers or, for reasons of load distribution, the requests can be distributed to several computers.

The data which the selection component 121 receives from the user device 11 , such as the position specification or the user identification, can be entered by the user via the input component 113 . B. is implemented as a keyboard or as a speech recognition module. In other implementation variants, this data can also be partially determined or supplemented automatically by the system, e.g. B. on the position in the form of identifiers of radio cells (Cell-ID) or one or more radio masts on which the mobile phone is logged in, or which are known to the mobile phone, or by means of data of a GPS position determination component with the User device is connected, or is in it.

The input data can either come directly from the user device 11 or from the network 13 , 14 , 15 , possibly also from other computers, e.g. B. a speech recognition software on a computer that answers calls. The user identification can thus have been entered by the user or it is defined by the telephone number or by data which are located on the so-called SIM card in the mobile phone or as a combination of these possibilities. The essential performance of the invention, namely the efficient selection and presentation of the information, is not affected.

In FIG. 2, a method according to an embodiment of the invention illustrated in an overview. The method is divided into two interacting processes, namely a search process 21 and a system process 23 . In the search process 21 , on the basis of input data (location, time, user ID), a lot of information offers are generated which come from the information offer database 123 ( FIG. 1). In the system process 23 , the internal data structure of the system, which influences the search process 21 , is modified on the basis of input data (location, time, user ID and the user's selection of offers).

The search process 21 proceeds in the following three stages, input data in the present exemplary embodiment being location, time and user ID.

As a first stage, based on the location and time information, a static preselection 211 is made from the existing database of the information offering database 123 ( FIG. 1). It is filtered out of the total amount of offers, an offer quantity which, according to the location information, belongs to a geographic area and can be used at the time of the inquiry, e.g. B. Open shops.

In the second stage, at least one user profile and user group profile assigned to the user is loaded based on the user ID. The profiles contain the statistical data and rules that are partially generated by the system process 23 . By evaluating 212 the statistics and rules, the offers of the offer quantity are evaluated, prioritized and possibly removed.

The offer quantity is structured and graphically prepared in the third stage 213 in order to be able to display it according to the physical conditions of the user device 11 ( FIG. 1).

He then selects an offer in step 22 . This is registered together with the input data used for the search in a selection report and then displayed.

The system process 23 takes place concurrently, the input being the selection protocol mentioned above. System process 23 consists of two concurrent or sequential sub-processes: statistics logging and rule generation. These processes generate the statistics and rules used in the second stage 212 of the search process 21 . Through this feedback, the invention can analyze the previous selection behavior and use it to determine the selection in future inquiries.

Fig. 3 shows an example of the method to illustrate the data of the information offer as a table. The offers 31 are listed , to which a time period 32 indicating the availability is assigned, for. B. the opening times and an address 34 of the offer. In other implementation variants, several periods and addresses can have been assigned to an offer.

The offer 31 has also been assigned to a geographical area (for example via the address). However, the area 31 can also have been assigned to several areas which, for. B. represent the sphere of activity of the offer. Each offer 31 is assigned to a category 33 , which describes the content of the offer 31 . Since category 33 can in turn be contained in other categories 33 , the path to the top category 33 of level 1 is shown here.

4 shows an example of a category graph to illustrate the method, which is used to categorize the offers. In this case, a tree representation is selected; in other variants, nodes could also have several higher-level nodes. In still other implementation variants, an offer could also be assigned to several categories. In general, the data could also come from different data sources and tables.

Fig. 5 shows an example of the contents of the geo-information database 123rd This assigns individual streets or street sections to a geographical area. These areas form a grid, as shown in FIG. 6. The data from FIG. 5 are used, on the one hand, to assign offers to geographic areas, e.g. B. about their address, on the other hand to assign position information to an area.

Fig. 7 shows the processes that are carried out during the selection of the offer. In the exemplary embodiment described here, a search query 71 contains, as input, a position specification, optionally one or more keywords, a user identification and the time of the query. The position specification and especially the keywords (and in some versions also the time) are optional. These are then only used in the static selection process 72 , 211 . In this process 72 , 211 , the amount of offers is limited as much as possible on the basis of static data by excluding irrelevant offers.

The quantity then left is further prioritized, filtered and possibly reduced in the dynamic selection process 73 , 212 based on statistical data and rules 792 from the databases 126 , 127 . The resulting quantity of offers is then prepared 74 in a user-friendly manner through a presentation process and compiled into answer pages which are specially designed for small display components 111 . This data is then made available to the user 75 as an answer and is displayed by the user device 11 . Steps 74 and 75 in FIG. 7 together correspond to step 213 in FIG. 2.

The user navigates through the display of the offer selection. When the user selects 76 an offer, the selection is registered 77 and stored together with the input data used in a selection log 791 . The selected offer is then shown to the user 78 . The user can then select further offers (steps 75 , 76 , 77 , 78 ) or start a new search (steps 71 , 72 , ... ).

The process of the selection protocol evaluation 790 is part of the system process 23 shown in FIG. 2. Process 790 reads log data 791 and generates statistics and rules 792 . Through this feedback, the invention can analyze the previous selection behavior and use it to determine the selection in future inquiries.

FIG. 8 shows the process of static selection (process 72 in FIG. 7 or 211 in FIG. 2). In the static selection, the number of offers is initially limited based on an item specification (step 81 ), then (if possible) by taking into account the desired categories (step 82 ) and finally by considering the time of the request (step 83 ).

Referring again to FIG. 5 and Fig. 6 directed. The regions of the geographic information database 124 illustrated in FIG. 5 correspond here to the smallest (so-called atomic) regions 61 that are no longer divisible ( FIG. 6). For the sake of simplicity, atomic areas are all of the same size and numbered as an example: numbers for the rows, letters for the columns. Other numbering types and territorial forms are also conceivable. Also, the areas do not all have to be uniform.

The areas assigned to the offer data in the database 123 are variable in size and consist of several adjoining atomic areas, as shown in FIG. 6. Areas assigned to the offers are called variable areas 62 here. Each atomic region 61 is also assigned to exactly one variable region 62 . So there is no overlap. The aim is that each variable area is assigned only a maximum number of offers if possible. e.g. B. 400. If this upper limit is exceeded by adding offers, then the variable area 62 is divided into a plurality of variable partial areas. e.g. B. in four areas. The offers are then assigned to one of these new sub-areas. The division can only take place until the variable area 62 is an atomic area, ie has not reached the minimum area size.

In this way, small areas are created in geographical areas with a high supply density and large areas in those with a low density, as shown in FIG. 6. When removing offers, the same procedure can be followed and several adjoining variable areas can be merged into one variable area as long as the upper limit is not reached.

The process for selection based on the position specification shown in FIG. 9 is carried out if a position specification (e.g. location and street name) exists (condition 91 ). The variable areas associated with the position specification are then determined (step 92 ). For this purpose, the position information is mapped to an atomic area with the aid of the geographic information database 124 (cf. also FIG. 5). Depending on the form of the position specification, it may also be possible to use differently structured geographic information databases or algorithms (GSM cell identifiers of radio masts, geocoordinates in various forms, etc.).

The atomic area determined in step 92 is located within a single variable area. As a result, the offers associated with the determined variable area are selected from the total amount of offers (step 93 ). The offers from neighboring variable areas may also have to be taken into account, e.g. B. if the variable area is too small or the atomic area of the position is at the edge of the variable area. Optionally, the selected offers of the adjacent areas considered in this way can be rated lower in a subsequent evaluation process. In this way, the selection of offers is drastically restricted and location-relevant offers can be quickly determined from the total offer quantity.

If no position is specified, the "No" branch is selected in condition 91 . In this case, all offers are location-relevant.

FIG. 10 shows the reduction in the quantity of offers on the basis of keywords, if these are present. Keywords can be used as input parameters to further restrict the quantity of offers. In general, an attempt will be made to automatically determine the selection parameters such as position information or user identification, so that the user does not need to make any further information. In individual cases, however. B. if the displayed amount of offers is not specific enough or the user wants to express a special request, the user could specify one or more keywords as search terms, select or tell a speech recognition component, e.g. B. "Taxi".

If there are keywords, the "yes" branch of condition 101 is selected. Each key word is then associated with one or more categories (step 102 ), for example when a key word completely matches a category name or with the prefix of a category name. Another option is to evaluate the selection protocol and determine which categories, how often and with which key words were ultimately chosen. This means that keywords that do not appear in the category graph can also be assigned to categories. The probability of the relevance of a category for a certain keyword can be determined.

After the categories have been determined from the keywords in step 102, those that are not assigned to any of the determined categories are sorted out from the available quantity of location-relevant offers in step 103 . If there are several keywords, an "or" link is assumed here. Other search queries with more complex bracketed expressions, with "and" or "not" operators are conceivable.

If there are no keywords, then the "No" branch of condition 101 is selected and no offers are sorted out.

FIG. 11 illustrates the process of selection based on the date of the request. Offers that are relevant for the time of the request are selected from the amount of information available after the previous selections. If no time for which a search is to be carried out was explicitly specified in the request, then the time of the request is used implicitly. This is determined via the time of the clock 125 of the computer 12 .

If the request time was determined using the clock 125 , then the time may have to be corrected. Since it is about offers z. B. to select according to the opening times, the time at the location of the user device is decisive, not the time at the location of the computer 12 . If there is a position information in the search query (“yes” branch of query 111 ), the point in time can be corrected in step 112 . For this purpose, the position is assigned to an atomic area. Since a generally known time zone can be assigned to each area, it is possible to determine in which time zone the user device is located. Relative to the location of the computer that makes the selection, this results in a time difference z. B. -1 hour. As a result, the time of the request can be transformed at a time that corresponds to the time of the request from the user device 11 . Does the request meet z. B. at 14:00, then this corresponds to a search query that was sent to user device 11 on site at 13:00 h. The transmission time is neglected.

If the time was corrected or no position was specified or the time was explicitly specified, then the selection of the offers based on the time is continued in step 113 .

As described, information offers are each assigned one or more time periods, e.g. B. Opening times. In step 113, the offers are selected for which the point in time lies within a time period which is assigned to the offer. Realization variants with other rules are also conceivable, e.g. B. a time period that is marked so that the time must be outside the time period. In preferred implementation variants, the point in time will not have to be exactly in the time span, since distances to be covered must also be taken into account. An offer that is only open for 1 minute but is so far away that it cannot be reached in a minute is rated low. An offer that is not yet open, but may be open in 5 minutes, will still be considered in the selection.

Although here the selection by position, keyword and time is sequential has been described is a different order or a parallel execution of the Selection also possible. In a preferred implementation variant, the individual criteria combined into a single criterion and thus the Selection made. In other variants, individual static selections can be made processes such as B. completely omitted the selection with the help of keywords become.  

In another variant, before the offers are evaluated on the basis of statistical data when the number of offers falls below a defined minimum number, it can be defined that the static selection is to be repeated based on the position information, with a larger number of variable areas, and thus also more distant offers included in the quantity (possibly until the specified minimum number of selected offers has been reached). The selection by evaluating statistical data is based on data that are obtained from the logging of selections already made by users. The logged data is stored in the log file 791 ( FIG. 7). In the log evaluation process 790 , the log file 791 is evaluated and the data are aggregated into statistics, which are e.g. For example, indicate the likelihood of a particular offer being selected at a particular location. With the help of these statistics, the offers of the previously selected offer quantity (through static pruning) are evaluated.

In addition to the statistical data, preferred implementation variants in addition, rules for evaluation are also used. Make rules Statements such as B. "between 12:00 and 14:00 in the area A2 preferably an offer in the category restaurant selected ". Rules can be determined from the statistical data are obtained or rules are explicitly defined, e.g. B. from user: "User Müller doesn't like Chinese". Rules that determined Exclude offers will reduce the quantity of offers beforehand evaluated.

The log file 791 , a proposal for a specific aggregation of the log data to statistical data and a representation of rules are described below as an example of an implementation variant . Other aggregation variants and / or other representations of rules are provided in alternative embodiments.

After the user is presented with a selection of offers as a result of the search query, the user will select one or more offers. The search criteria used and the ultimately selected offers are also logged in the log file 791 . The content of such a log file is shown by way of example in FIG. 12.

The data in a log line usually comes from two requests, namely the original search query and the request for details for the selected one Offer. This data is merged. One line therefore contains that for Query relevant input parameters: the identification of the user currently set user profile, the position z. B. as an atomic area or place and street, possibly one or more keywords and the time the request.

If the selected offer quantity is presented and the user now selects an offer, the line is supplemented by the user selection data: z. B. the time when the selection was made, the offer and the category in which the offer was. The supplemented line is then entered in the log file 791 . If the user selects an offer again from the quantity already presented, a new line is generated, the input parameters relevant to the search being repeated and the new user selection data being added. One line is created for each selection process.

Although only one log file 791 is shown here, it is also possible to use a plurality of files which only record the user behavior for a specific period of time, or only individual aspects thereof.

In the exemplary embodiment described here, the log file 791 explicitly contains a user profile. User profiles are based on user groups. These groups can be formed according to various criteria, e.g. B. demographic, regional criteria, or behavioral similarities. Examples of groups are "businessman" and "tourist". A user can in principle belong to different user groups at the same time or by specifying rules depending on the time different z. B. from 6:00 pm "Meier Privat" profile. The definition of a user group and the affiliation to a user group can also result implicitly from the evaluation of user and log data. For the sake of simplicity, only the explicit affiliation to a single profile is considered here, the user himself being able to use rules to define which profile he is assigned to.

The logging of user behavior, the subsequent generation of the statistical data and rules as well as the use of this data for the evaluators The relevance of offers in the future results from feedback Self-adapting system that learns from previous user behavior. This is based on the assumption that a selection behavior of a user or a user group in the near past the selection behavior in the near future will be similar.

Statistical data can be generated from the log data according to various aspects by summary. The following types of statistical data and rules can be distinguished:

• - Statistical data and rules for users: These are needed so the system remembers user-specific behavior and this knowledge used for requests from this user to evaluate the offers.
• - Statistical data and rules for user profiles: This takes into account System behavior of entire user groups and includes them user-group-specific profiles.
• - Statistical data and rules for areas: The system takes into account Similar behavior of users or user groups on one specific place.
• - Statistical data and rules for offers: These give e.g. B. which ones User groups have used a service. These data are more for Information of the providers of the offers to choose from.

Fig. 13 shows an example of how user-specific can be recorded, which offer and which offer category in which area and how often were selected. The data is generated from the log as follows. When a person selects an offer in an area for the first time, an entry is created both for the offer itself and for each category that is assigned to the offer in the log. If there is a line in the log in which the same user chooses the category again, the counter of the frequency of calls is counted up.

This then indicates how often the category chosen by the person at the location has been.

For administrative reasons z. B. fixge the time of the last change to remove entries that have not been used for a long time can.

In this example, only the frequency is determined. That would be the truth Determine the probability of a call using the total number of entries. But in the preferred implementation variant, the absolute frequency is sufficient, because the total number of entries changes very often.

Depending on the implementation variant, the data can be aggregated differently. Examples are:

• - How often which category or which offer from which person or user group has been selected.
• - How often which category or offer from which person or User group in which area was selected.
• - How often which category or offer from which person or User group in which area was selected at what time.
• - How often which category or offer from which person or User group at what time was selected.

In all of these example cases, entries can e.g. B. according to different Criteria such as time of day: "Morning", "Evening", "12: 00-14: 00", or days: "Mon-Fri", "Sat, Sun", or seasons can be summarized.

Statistics can be summarized independently of the search process, e.g. B. at fixed intervals or with a sufficient amount of new log data.  

Fig. 14 shows an example of user-specific rules in a table. The table shows that BorgerKong is very popular with user "Hugo", Maccie is not very popular and Chinese should not be offered at all.

In general, rules can be determined from statistical data by grouping, on the other hand, they can also be specified explicitly. The advantage of explicit rules lies in readability for users. Users can e.g. B. Define exclusion rules that are not derived from statistical data can.

Rules can not only be user-specific, but of course also profile, location, Time and category specific.

In the exemplary embodiment described here, rules are used in addition to statistical ones Data held and evaluated. In other variants, rules can be dispensed with be, or they are generated from the statistical data and serve the Modification of the statistical data.

The statistical data and rules are used to evaluate offers and their categories in a lot. However, rules can also be used to exclude Offers are used, which is checked before the evaluation.

A rule applies to an offer if the offer or category matches the Information in the considered rule matches. If a period of time in the Rule is present, then the corrected time of the search must also be question lie within the time span. If a position in the search request exists and an area was specified in the rule, then must the position is in the area.

First, the rules are applied that exclude offers or Categories serve, then the frequencies from the different states weighted by multiplication by a weighting factor and then added or if probabilities instead of frequencies used, multiplied.  

The rules that are not used for exclusion are evaluated by Anga ben how "high" and "low" are converted into multiplication factors. Meets one Usually towards an offer, then the previous evaluation will correspond to that multiplied the factor. Variants are also conceivable in which instead of multiplications fixed factors are added.

If there are more offers than a certain upper limit (e.g. 50), lead only the highest rated offers and will be included in the selection. The quantity of offers determined in this way becomes the presentation process to hand over.

The above mentioned limit for the number of offers can also be from the user zer have been specified if this z. B. explicitly a larger supply quantity want to get as the end result. Depending on the implementation variant, other Criteria are used. So z. B. an assessment is dispensed with, if the number of offers falls below a lower limit (e.g. 3). Other Exclusion criteria can be minimum limits for the assessment of an offer be botes. If this limit is not reached by the offer, then it will removed from the result set. The statistical Data and rules, as described above, take place sequentially or in parallel. Similarly, the criteria such as the maximum upper limit after each evaluation being checked. Further optimization can also be done after each evaluation it is determined whether the evaluation of the currently evaluated offer in the Relationship to those already rated is so bad that it probably isn't will show up in the end result.

Since the evaluation can be complex and therefore lengthy, in some cases it is over management variants provided to carry out the evaluation in stages, the Statistics and rules with a large weighting are used first. At a sufficient differentiation between the offers or the over If a time limit is exceeded, the evaluation can be terminated prematurely.

The end result of this selection stage results in a number of evaluated offers, as shown by way of example in FIG. 15. The individual ratings of the categories and offers are combined to form an overall rating such that a higher rating of the category leads to a higher overall rating, e.g. B. by adding or multiplying the individual ratings by a weighting factor. If there is no rating for an offer or a category, an average value can be accepted for rating.

The amount of the evaluated offers is divided by the presentation process (e.g. step 74 in FIG. 7) into several so-called pages (e.g. in WML format), which can be linked together to form a tree, as shown in FIG. 17 shows exemplary. The display component 111 in the user device 11 represents a single page. An entry can then be used to navigate to a connected page or to select an offer. The division takes into account the limited display capacity and enables quick navigation to highly rated offers.

FIG. 16 describes the method of transforming the evaluated list into a tree consisting of several pages, as shown by way of example in FIG. 15.

The root of this tree is a page with a "high score" area, in which a certain number (e.g. 3) of the highest rated offers are displayed (area 171 in FIG. 17). This page is supplemented below by proceeding as follows with the list of offers. First, the categories of the first level are sorted in descending order according to their respective rating (e.g. transport ( 50 ), business ( 50 ), hotel ( 10 ), food ( 10 ), cinema ( 10 )). Then proceed for each category, starting with the highest rated, as follows: If a total of less than a certain upper limit of offers (e.g. 5) belong to this category, then all offers go directly to the current page without category information if the page does not exceed a maximum size. The entered offers may be sorted according to their rating.

If there are more offers than specified by the upper limit in a category (e.g. more than 5), then a reference to a new page is entered in the current page (e.g. "Hotel"; reference number 172 in FIG. 17 ). The new page and the offers are processed recursively as described above. So a new page is created for the category (e.g. for the hotels) and the categories of the second level are sorted in descending order according to their respective rating. (3 stars ( 100 ), 4 stars ( 50 ), 5 stars ( 10 ),...).

If the number of entries exceeded the maximum page size, then a reference is generated which points to a successor page (eg "More..."; Reference character 172 in FIG. 17) and continues with the compilation of this page.

In order to optimize the input speed on mobile user devices with a small screen and limited input options, the following input methods are used in the exemplary embodiment described here:
If entries are required that consist of several elements, the individual elements can be separated in a single input field by delimiters (e.g. "-", "/" or "_") and entered without a prescribed sequence. If, for example, the elements "country", "city", "street" intersection "street" are to be entered for an exact position determination, then the entries are not made as usual in four separate text fields, but only in one, e.g. B. "Germany-Munich-Luisenstr-Elisenstr". The order doesn't matter. The algorithm first divides the input into individual elements using the separators, then assigns one or more types to each element using a database, in the example "country-city-street-street". Based on the types, the elements are brought into a standard order and compared with a database of possible entries.

The entry of elements is still allowed. B. for the position information on Common and often used standard abbreviations from other fields of application ten, e.g. B. Countries and place names of license plates or vehicle country codes Chen, air traffic, sporting events u. Is an abbreviation not known or ambiguous, the user is presented with a selection or clarification dialog. The system can create new abbreviations by evaluating user input learn.

Many mobile devices do not have a complete one to save space alphanumeric keyboard. Instead, the device has a limited amount on input keys, whereby one key is assigned to one menu by multiple assignments is marked as responsible by characters, e.g. B. "abc" and from this set  again a character is particularly distinguished, e.g. B. "a". That especially out Drawn characters can be easily generated by pressing the button while generating the non-distinguished characters (e.g. "b" and "c") further actions are necessary, e.g. B. simultaneous pressing a special Multiple assignment selection key (e.g. SHIFT) or pressing the same key in a row at a short distance, e.g. B. pressing twice Key "a" gives "b", pressing "c" three times.

Fig. 18 shows the portion of the input unit of a mobile phone. For example, to enter the "LUISENSTRASSE" parameter, press the following keys: "555 88 444 7777 33 66 7777 8 777 2 7777 7777 33".

In order to optimize the input of information by means of this type of input device, it is possible to use only the excellent characters for the input, instead of the complicated to enter not marked characters of the multiple assignment quantity of the respective input key, as z. B. instead of a "b" or "c" enter an "a". This initially creates a misrepresentation due to the input device (for the example "LUISENSTRASSE" the representation "JTGPDMPTPAPPD"), which can, however, be easily corrected by the computer receiving the input if there is only a limited amount of possible and meaningful inputs, e.g. B. the set of all street names of a city. The receiving computer 12 has stored in a database in addition to the possible and meaningful inputs (for example "LUISENSTRASSE") and their misrepresentation ("JTGPDMPTPAPPD"). By allowing the misrepresentation, it can happen that more than one possible entry has the same misrepresentation. In this case the user is shown a selection dialog from which he can choose between the different meanings.

For the example above, the number of actuators is reduced Use this procedure to move from 35 to 13.

For optimized input of cardinal points or neighborhoods on mobile input devices with a numerical keyboard (or a numeric keypad as part of input component 113 ), the standardized arrangement of the number keys can be used to enter cardinal points or geographical positions of neighborhoods by pressing a key by the arrangement of the keys on the number pad is associated with a very abstract map, or wind rose.

Fig. 18 a possible cut of the input unit is an example of a mobile phone. The following scheme results when entering cardinal points: 2 = north, 3 = north-east, 6 = east, 9 = south-east, 8 = south, 7 = south-west, 4 = west, 1 = north-west. Similarly, the different parts of a city can be abbreviated with the additional meaning 5 = center.

The highest input efficiency is described by the combination of all Single procedure reached. So the exact position information "Country: Germany, City: Munich center, street: Luisenstrasse, corner: Elisenstrasse "through the Misrepresentation "D-MJ-DJGP-JTGP" from "D-M5-ELIS-LUIS" on a mobile phone with just 14 keystrokes.

Claims (16)

1. Procedure for making a pre-selection of information offers, with the steps:

• a) receiving a request which contains at least location information and a user identification,
• b) accessing stored behavior information relating to the user identified by the user identification,
• c) accessing an information offering database ( 123 ) which contains data about a large number of information offering, said data containing at least location information and category information for each information offering,
• d) Evaluating several information offers from the information offer database ( 123 ), for each information offer to be assessed depending on at least the location information and the category information on the one hand and at least the location information and the behavioral information on the other hand, to what extent the information offer meets the presumed wishes of the Corresponds to the user, and
• e) preselection of at least one information offering from the information offering database ( 123 ) in accordance with the evaluation determined in step d).

2. The method of claim 1, wherein the request is also a request time assigned in step d) when evaluating the multiple information offers with regard to the presumed wishes of the user is taken into account. 3. The method according to claim 2, in which the evaluation with regard to the request time is carried out by comparing the request time with a time period assigned to the information offer via the data in the information offer database ( 123 ) or via the behavioral information. 4. The method according to any one of claims 1 to 3, wherein the behavioral mations on the previous selection behavior of the user and / or on the previous selection behavior of at least one user group of the user are based.   5. The method of claim 4, wherein the behavioral information about the previous selection behavior of the user or user group in the form of Statistics and / or records and / or rules are available. 6. The method according to claim 4 or claim 5, with the further steps:

• a) outputting the at least one information item preselected in step e) for display to the user,
• b) determining an actual selection of the user from the at least one information item preselected in step e), and
• c) updating the behavioral information depending on the actual selection of the user determined in step g).

7. The method according to any one of claims 1 to 6, wherein the behavioral mations on a selection behavior of the User and / or based on a selection behavior in the form of rules based on at least one user group of the user. 8. The method according to any one of claims 1 to 7, further comprising the request at least one keyword is transmitted, which in step d) during the evaluation of the several information offers regarding the presumed wishes of the User is taken into account. 9. The method according to any one of claims 1 to 8, in which for evaluating the information offers from the information offer database ( 123 ) with regard to the location information and the location information, the information offers are grouped into geographically oriented areas, wherein when a first predetermined number of offers in an area is exceeded this area is divided, and if the value falls below a second predetermined number of offers in an area, this area is combined with neighboring areas. 10. The method according to any one of claims 1 to 9, which is executed by a computer ( 12 ) which can exchange data with a mobile user ( 11 ) via a network ( 14 ). 11. The method according to claim 6 and claim 10, in which in step f) the at least one preselected information offer is transmitted via the network ( 14 ) to the mobile user device ( 11 ), and in which the data thus transmitted is transmitted by the mobile user device ( 11 ) be prepared for a presentation on a display ( 111 ) of the mobile user device ( 11 ) with limited display options. 12. The method according to claim 10 or claim 11, wherein the network ( 14 ) comprises at least one GSM, GPRS or UMTS network section. 13. The method according to any one of claims 10 to 12, in which the output data a GPS receiver used to determine the location information become. 14. Computer ( 12 ), in particular Internet server, which is designed to carry out a method with the features of one of claims 1 to 8. 15. Mobile user device ( 11 ) which is designed to carry out a method according to one of claims 11 to 13. 16. Mobile user device ( 11 ) according to claim 15, which is designed as a mobile phone, electronic organizer, PDA, laptop, notebook, handheld or PC.