DE102014215266A1 - A method for providing a factual information relevant to a user of a personal digital assistant, personal digital assistant and storage medium - Google Patents

Abstract

In a method for providing a factual information relevant to a user of a personal digital assistant, the provision of user-relevant information with regard to the selection of such information is improved by a normality value in dependence on past context information (100) and current context information (100 ) is determined, depending on the normality value decided by a classification of the factual information as relevant factual information, the factual information is provided only when grading as relevant factual information.

Description

The present invention relates to a method for providing a factual information relevant to a user of a personal digital assistant, a personal digital assistant and a storage medium with computer-readable code for controlling a personal digital assistant for carrying out the method according to the invention.

Personal digital assistants, also referred to below by the abbreviation PDA, have gained considerable importance for everyday use. In the context of the present invention, a PDA is understood as any digital mobile device that can be assigned to a user personally and with which the user can obtain relevant information for him / her. Personal digital assistants are usually equipped with numerous technical capabilities, in particular satellite navigation (eg GPS), mobile communications (eg GSM, UMTS, LTE, WLAN, Bluetooth, NFC), one or more cameras, often designed as touch screen display panels, speakers, powerful Microprocessors and volatile and non-volatile data storage. Due to the large number of sensors as well as the possibilities for data storage, transmission and processing, a PDA can record, store and process data for its user in a far-reaching manner or transmit it to an external computer for storage and / or processing. Examples of personal digital assistants include in particular so-called smartphones. A PDA within the meaning of the present invention may also be the so-called infotainment system of a motor vehicle, which may have, for example, a navigation system, a touchscreen, various output devices, a voice control, a mobile radio unit and a computer. However, such an infotainment system must be able to be personally assigned to a user, which is often possible in modern motor vehicles. This avoids mixing data from different users. This does not mean that a PDA can only be used by a single user. When used by multiple users, however, it is necessary that the respective current user is known to the PDA, so that the PDA correctly assigned information or can use already existing information correctly. The current user can be made known to the PDA, for example, by a user login.

US 2013/0345961 A1 describes a method for providing route suggestions of a mobile device. It identifies destinations and routes based on past behavior patterns of the user and a current context that the user is likely to choose. Taking into account foreseeable events concerning a journey on the identified routes, a route proposal is provided to the user. Past behavior patterns include places visited by the user in the past, the times when he did so, and means of transport that the user has chosen. The current context includes, in particular, the current geographical position of the user as well as entries in his calendar. The mobile device thus provides the user at all times with information about the situations that are most likely to arrive.

According to the methods known in the art, supporting information is offered to the user at all times. This is particularly helpful for the user when he is in a situation unfamiliar to him or is about to embark on such a situation. For example, if the user is located in a location he or she is unfamiliar with, he or she will be assisted in selecting destinations (e.g., restaurants or hotels) and routes there. However, as a rule, a user is in familiar environments and situations. The user is then not dependent on the supporting information and may find it annoying and may even be distracting and disturbing.

It is therefore an object to improve the provision of user-relevant information with regard to the selection of such information.

The object is achieved in a method having the features of claim 1 and in a personal digital assistant according to claim 9. Advantageous developments of the invention are subject matters of the dependent claims. The inventive idea is particularly suitable for controlling a computer designed as a PDA, which is why the invention can also be implemented in whole or in part in the form of a computer program product which, when loaded and executed on a computer, completely or partially effects a sequence according to the invention. It can be provided, for example, in the form of a storage medium according to claim 10.

• – Bestimmen einer den gegenwärtigen und / oder einen zukünftigen Zeitpunkt betreffenden aktuellen Kontextinformation,
• – Laden einer mindestens einen vergangenen Zeitpunkt betreffenden früheren Kontextinformation und
• – Auswählen einer Sachinformation in Abhängigkeit der aktuellen Kontextinformation und der früheren Kontextinformation.

A generic method comprises the following steps:

• Determining a current context information concerning the present and / or a future time,
• - Loading a previous context information relating to at least one past time and
• Selecting a factual information in dependence on the current context information and the previous context information.

In this way, such factual information is selected that relates to the current context of the user. This can also be done depending on the previous context information, such as taking into account previous habits of the user when predicting a user-selected destination. In other words, information about likely visited locations can be obtained from statistical models via the e.g. From the previous movement data known habits of the user can be obtained. An example of factual information that does not affect the user's current context is restaurant recommendations for a specific city in which the user will not be or will be present or in the foreseeable future. Such factual information is therefore not selected by a generic method. On the other hand, factual information concerning the current context of the user is basically selected. This happens even if the user is well-versed in such factual information, for example, because the factual information relates to a regularly traveled by the user route or a regular restaurant of the user near his apartment. Such factual information is not relevant to the user within the meaning of the present invention.

According to the invention, further method steps are carried out in order to decide whether the factual information selected in this way is to be provided to the user or not. In one step, a normality value is determined as a function of the past context information and the current context information. The normality value is a measure of the normality of the current context of the user. The normality value is determined by comparing current and past context information. For example, if this information differs greatly from one another, it can be concluded that the current context of the user is very different from past contexts. In other words, the user is in a situation unfamiliar to him. The normality value reflects this by taking, for example, a small numerical value or the logical state zero (0). On the other hand, if the current context information is very similar to the previous context information, then the user is presumably in a familiar situation. The normality value then assumes, for example, a large numerical value or the logic state one (1).

Depending on the normality value, it is decided in a further step according to the invention whether the factual information is classified as relevant factual information. In this case, the factual information is classified as relevant if a normality value reflecting a low degree of normality of the current context of the user has been determined. Conversely, the factual information is not classified as relevant factual information if the normality value indicates a high degree of normality of the current context of the user.

Finally, according to the invention, the factual information is only provided if it has been classified as relevant factual information. In other words, the invention spares the user such information that concerns his current context but in which he is not interested because of the normality of his current context because the information is already known to him or he can easily access it without support ,

In a first advantageous embodiment of the invention, the current context information and / or the earlier context information relate to a location, a path and / or an event. Location information may include a geographic location, a time associated with a location, and / or a sequence of location information. A route information may be a route defined by geographical location information, a travel modality (eg, car, train, plane, bicycle, foot, etc.), a travel pattern (eg, a sequence of successively traveled routes), and / or times associated with travel distances (eg, departure time and / or arrival time). Event information may include a time of an event (eg, time of waking up, time of leaving a home, start of work, etc.) and / or the occurrence of an event (eg, a refueling process, the invocation of a particular smartphone app, the use of a navigation system , a phone call to a specific person, etc.). Other information such as weather information, traffic information, a television program, a movie program, etc. may also be included in context information. Context information may include information that is not as such personal information of the user. However, they must relate to the personal context of the user. For example, weather information or Traffic information no personal information of the user. However, weather information about a user's location or traffic information about a user's traffic route relates to the personal context of the user and therefore may be included in context information.

The individual information encompassed by a current and / or earlier context information can be obtained by the PDA in known manner from personal and / or public sources as well as by own means. Current location and time information can be determined by the PDA, for example, by means of its satellite navigation unit or its internal clock. Information about places likely to be visited by the user, the times associated with them, and other time-related events can be found in the user's calendar, which can be e.g. stored on the PDA itself or on a server accessible via data connections. Weather and traffic information can be obtained from servers where this information is provided by service providers for retrieval.

The advantage of the mentioned embodiment results from the knowledge of the inventors that the normality of a situation largely depends on the information of a location, a path and / or an event. This information significantly characterizes the behavior of a user. Therefore, with the help of this information, the normality of a current context can be assessed with high reliability and at the same time with little effort.

With further advantage, the step of determining the normality value in dependence on the past context information and the current context information comprises the following individual steps. In a first single step, a statistical parameter is determined as a function of the previous context information.

The statistical characteristic may in particular comprise a frequency distribution. For this purpose, similar information from the previous context information is first selected. The frequency of this similar information is detected with reference to a discrete-value and / or time-discrete parameter space. The detection is also conceivable with reference to a value- and / or time-continuous parameter space, but it is expedient to discretize such a space appropriately first. Graphical representations of such frequency distributions are also called histogram or "heat map". The term similar information in this context refers to such information that can be represented both in the common parameter space as well as content related. For the determination, all similar information of the earlier context information can be selected. However, it is also possible to select only a subset of similar information from the previous context information. An example of similar information is where the user was. In this example, a suitable parameter space can be given by the two parameters longitude longitude and latitude latitude (each divided continuously or into value discrete value ranges). In another example, the similar information is given by times when the same event occurred. These can be, for example, the times of a morning getting up. In this example, a suitable parameter space can be specified by the parameter of the time (again continuous or time-discrete). Learning methods for estimating frequency distributions are known in the art. An example of a suitable learning method is known as "expectation maximization".

It may be known in advance that the frequency distribution can be described by a normal distribution (Gaussian distribution). This can be assumed on the basis of the central limit theorem, for example at times which fluctuate around an average due to a multiplicity of statistically independent influences. For example, the travel time required for a particular consistent route is approximately normally distributed, as experience has shown. The same applies, assuming a constant departure time, for the corresponding time of arrival.

A more complex frequency distribution may result when the frequencies of several similar information are summarized (e.g., all departure times or all arrival times of a day). The resulting frequency distribution can be described as a mixed distribution or a composite distribution. If the individual distributions forming the mixed distribution are normal distributions, the mixed distribution can be described as a Gaussian Mixture Model (GMM).

It should be noted that the determination of a frequency distribution is not limited to parameter spaces defined by physical units. It is known to determine a frequency distribution over other parameter spaces. For example, the frequency of visited places can also be higher than that Parameter space containing the names of all visited places as strings. Methods for determining a frequency distribution are also known in the art as clustering methods. Examples of clustering methods suitable for the practice of the invention include density-based clustering techniques. Such a method has become known by the technical article "Discovering Personal Gazetteers: An Interactive Clustering Approach" by C. Zhou et al., Published in "Proceedings of the 12th Annual ACM International Workshop on Geographic Information Systems," pp. 266-273, New York, NY, USA, 2004 , Examples of clustering methods suitable for the practice of the invention further include different embodiments of so-called "Spatio-Temporal Clustering". Such a method is known, for example, from the technical article "A Clustering-based Approach to Discovering Interesting Places in Trajectories" by AT Palma et al., Published in "Proceedings of the 2008 ACM Symposium on Applied Computing," pp. 863-868, Fortaleza, Ceara, Brazil, 2008 ,

• • H für Zuhause (englisch home);
• • W für Arbeit (englisch work);
• • R für Restaurant.

The statistic may also include an average of past context information. The term "mean" is not strictly mathematical, but is more commonly used to refer to an expression that appropriately summarizes the prior context information. In particular, an average must not be numeric, but may also be alphanumeric. For example, habitually searched by the user in a particular order places searched coded and used as a statistical index. For example, the following encoding could be used:

• • H for home (english home);
• • W for work;
• • R for restaurant.

If, for example, From the previous context information, it appears that the user habitually visits these locations in the order "H W R W H", so this string can be used as a statistical parameter.

In a second step, the normality value is determined as a function of the statistical parameter and the current context information. This determination can be made in different ways depending on the type of similar information used, the parameter space and the resulting statistical characteristic.

If the statistical characteristic comprises a frequency distribution, the normality value can be determined depending on the location of the similar information of the current context information in the frequency distribution. For example, the normality value may be determined as the frequency corresponding to the current context information. If the frequency distribution shows relative frequencies, the result is a normality value between zero and one.

If the frequency distribution is a normal distribution or the frequency distribution can be described by a normal distribution, then the normality value can be calculated as the sum of all frequencies which are even farther from the center of the frequency distribution than the current context information. If the current context information is e.g. an arrival time which corresponds exactly to the mean value of the frequency distribution normally distributed in a good approximation, the result is a normality value of one.

If the statistical parameter comprises an average value and in particular an alphanumeric mean value, then the normality value can be determined by means of the distance of the similar information of the current context information from the statistical parameter. Suitable distance dimensions are known in the art. The Levenshtein distance (also editing distance, English "Levenshtein distance" or "Generalized Levenshtein similarity") is an example of a suitable distance measure. If, for example, "HWRWH" results as the mean value as explained above and if the similar information of the current context information is also "HWRWH" (ie the user has spent a perfectly ordinary day in relation to the places visited by him), this results in a Levenshtein distance zero, from which, for example, a normality value of one can be determined. It is also possible to use more complex measures that can take into account more complex statistical dependencies.

In one embodiment of the invention, the current context information comprises at least two single context information. Then, the step of determining the normality value in accordance with the past context information and the current context information comprises the following individual steps. In a first step, individual normality values are determined for each of the at least two individual context information in dependence on the previous context information and the current context information. In this case, as in the above-described determination of a normality value, also in the determination of a Single normality value of the single context information similar information of the previous context information can be used. In a second step, the normality value is determined as a function of the at least two individual normality values. This can be done by a simple averaging. Preferably, the normality value is determined by a weighted average. The weights assigned to each individual normality score may weight or weight certain circumstances that arise from the user's current context or known preferences. If, for example, a user has an excellent sense of orientation, but a poor vision in the dark, deviations of the location or of the path can be underweighted in the determination of the normality value by averaging, but deviations in the external light conditions can be overweighted. If the user is now on the road during a day on a route that is rarely used by him, the PDA does not classify and provide any information on the route as recommended for the route. On the other hand, if the same situation occurs at night, the driver desperately needs support. Due to the strongly weighted single normality value of the individual information "external light intensity", the PDA will classify and provide the factual information as relevant rather than relevant. In addition, those single normality values whose underlying single context information is known to be strong and reliable indicators of normality can be more weighted. For example, it is known that the time of getting up in the morning is a strong and reliable indicator of the presence or absence of a working day. If, for example, the user is at an unusually late time, an overweighting of the corresponding individual normality value can ensure that useful factual information is provided to the user on this (probably unusual) day.

In a further embodiment of the method, the normality value is a real number between zero and a constant upper limit, and particularly preferably a number between zero and one. A normality value of zero then indicates current context information that is highly unusual for the user. For example, this may be a location where the user has never been. By contrast, a normality value of one indicates completely ordinary current context information, e.g. the residence of the user at his place of residence. This determination of the normality value not only allows an intuitive understanding, but also facilitates the further processing of the normality value, for example when averaging individual individual normality values.

With further advantage, such a normality value can be used to design the step of deciding on a classification of the factual information as relevant factual information as a function of the normality value. For this purpose, the normality value is first compared with a constant threshold value. This is particularly advantageous between zero and the constant upper limit. Thus, if the normality value moves between zero and one, then the constant threshold value is particularly advantageous at 0.5. In a second step, the factual information is classified as relevant factual information if the normality value is smaller than the threshold value. Otherwise, i. if the normality value is greater than the threshold, the factual information is not classified as relevant factual information. In simple terms, factual information can be provided to the user if and only if his current context is abnormal rather than normal.

In a further advantageous embodiment of the method according to the invention, the step of providing the factual information comprises a consideration of the normality value. The provision of information by the PDA can be done in different ways. The information may advantageously be communicated to the user by visual means (e.g., via a display), acoustically (e.g., via a loudspeaker), and / or by haptic means (e.g., via a vibration signal). In each way, the provision can be configured differently again, e.g. by displaying the factual information in the display differently sized and / or in different colors. According to the decision on a classification of a factual information as relevant factual information is binary. The factual information is either relevant or not. However, the relevance of factual information classified as relevant factual information may be variable. As a measure of relevance, the normality value can advantageously be used. If, for example, factual information has been classified as relevant because of a particularly low normality value (for example, close to zero), it can be assumed that the user has a particularly pronounced interest in this information item. Therefore, this information can be provided to the user very pronounced. This can e.g. through a large display of factual information on the PDA and a simultaneous beep.

• • Satellitennavigationsvorrichtung, z.B. für GPS und / oder Galileo,
• • Mobilfunkeinheit, z.B. für GSM, EDGE, GPRS, UMTS, HSPA und / oder LTE,
• • Vorrichtung für eine Kommunikation über drahtlose Netze, z.B. WLAN, Bluetooth und / oder NFC,
• • Kamera zur Aufnahme stehender und / oder bewegter Bilder.

A Personal Digital Assistant suitable for carrying out the method has units for processing information, that is, a microprocessor, volatile and / or non-volatile Data storage. Advantageously, in a further embodiment, the PDA has one or more of the following sensor devices, with which a current context information can be detected:

• • satellite navigation device, eg for GPS and / or Galileo,
• • Mobile radio unit, eg for GSM, EDGE, GPRS, UMTS, HSPA and / or LTE,
• Device for communication via wireless networks, eg WLAN, Bluetooth and / or NFC,
• • Camera for taking still and / or moving pictures.

To issue relevant factual information, the PDA may more advantageously have visual, auditory and / or haptic output devices. In a further embodiment, the PDA also has one or more input devices such as keyboard, microphone and voice recognition unit. Input device and output device may also be combined into a device, e.g. as a touch screen.

According to the invention, the PDA has a determination unit for determining a normality value as a function of the past context information and the current context information, and a decision unit for deciding on a classification of the factual information as relevant factual information as a function of the normality value. The determination unit and the decision unit do not necessarily have to be designed as separate hardware units. They can also be formed by a microprocessor of the PDA, by being controlled by means of a computer program product according to the invention.

The invention also includes a computer-readable code storage medium for controlling a personal digital assistant to perform the described method. Furthermore, the computer readable code may be stored and transferable on one or more servers in the form of one or more files.

The invention will be explained below with reference to exemplary illustrations. Show it

1 a hierarchically structured overview of different context information and

2 exemplary time sequences of different days of a user.

It should be noted that the illustrated figures and the associated description are merely exemplary embodiments of the invention. In particular, representations of combinations of features in the figures and / or the description of the figures are not to be interpreted as meaning that the invention necessarily requires the realization of all mentioned features. Other embodiments of the invention may include fewer, more, and / or different features. The scope and disclosure of the invention will be apparent from the appended claims and the complete description.

1 shows in a structured representation examples for implementing the invention suitable context information 100 , This context information 100 include location information 110 , Route information 120 and / or event information 130 , location information 110 include geographical locations 111 , times assigned to a location 112 and / or a sequence 113 of location information. A way information 120 can be a route defined by geographical location, a travel modality 121 , a travel pattern 122 and / or times associated with itineraries 123 include. An event information 130 may be a time 131 an event and / or the occurrence of an event.

2 shows four exemplary time sequences of different days of a user. It should be noted that the reference numerals of 2 refer to times with which places can be associated. It designates about the reference number 221 a period of time the user spends in his home. The timeline 201 extends essentially over a whole day, whereby in the present example not all hours are shown, ie the presentation does not cover 24 hours.

The daily routine 220 represents a normal day of the user. This day the user starts in his home 221 , At the time 222 the user wakes up and gets up. Point of time 222 can be easily detected if the user uses the PDA as an alarm clock. Point of time 222 however, it can also be detected by the user moving the PDA after the PDA has lingered longer (namely, all night). Numerous other methods known in the art for detecting a waking up and / or getting up may be used, in particular methods for physiological monitoring of the user. From his home 221 the user enters a drive 223 to work by car 224 at. From There he later takes a ride 225 by public transport to a specific restaurant A 226 at. Then a ride follows 227 back to work by public transport 228 , At a later time follows a car ride 229 home 230 where the day ends.

The daily routine 240 represents a relatively ordinary day of the user, but the slight deviations from the normal daily routine 220 having. This day 240 the user starts in his home 241 , At the time 242 the user wakes up and gets up. This time is a little later than the time 222 , From his home 241 the user enters a drive 243 to work by car 244 at. From there he will ride later 245 by car to a certain restaurant A 246 at. Then a ride follows 247 back to work by car 248 , At a later time follows a car ride 249 home 250 where the day ends.

The daily routine 260 represents a day of vacation and therefore a rather unusual day of the user. This day 260 the user starts in his home 261 , He wakes up at a much later date 262 on as a working day. After leaving his home, the user enters a car drive 263 in the restaurant A 264 at. Then follows a car ride 265 in a cinema 266 , Later, a car ride follows 267 home 268 where the day ends.

The daily routine 280 represents a working day of the user, which is very unusual because the user is on a mission this day. The day starts in a hotel 281 where the user at the time 282 wakes up. He travels by car 283 to the office of a customer 284 , Later he drives by car 285 in a restaurant B, from there later by car 287 back to the customer's office 288 , From there, the user later enters a car ride 289 back to his hotel 290 where the day ends.

The daily routines 220 and 240 show days that are pretty normal for the user. He therefore has little interest in factual information, such as route recommendations for his journeys, because they are already familiar to him. Things are different in the days 260 and 280 , Here the user is receptive to factual information because he can use it well due to the low normality of these days.

The following table lists single normality values as they are in 1 displayed context information (columns of the table) for each of the in 2 daily routines (lines of the table) could be calculated. The top line of the table contains the in 1 used, the left column of the table contains the in 2 used reference numerals. The single normality values in this example are between zero (lowest possible normality) and one (maximum normality). The total normality values are listed in the right column headed "Normality Value". 110 111 112 113 120 121 122 123 130 131 normal value 220 1 1 1 1 0.99 1 1 0.97 0.85 0.85 0.97 240 1 1 1 1 0.77 0.5 1 0.8 0.85 0.85 0.88 260 0.47 1 0.2 0.2 0.23 0.5 0 0.2 0.2 0.2 0.33 280 0.3 0 0.9 0 0.57 1 0 0.7 0.9 0.9 0.50

In the present example, the normality values result from arithmetic averaging of all individual normality values. So no single normality value is over- or underweighted. For example, the normality value for the day becomes 220 by averaging the values in the columns 111 . 112 . 113 . 121 . 122 . 123 . 131 educated. The values in the columns 110 . 120 and 130 are in turn mean values, which are listed only for the sake of completeness.

It is explained in columns below how the exemplary individual normality values were determined.

In the column 111 become the location information 111 (latitude and longitude) of all current locations of the respective day (current context information) compared with the frequency distribution of previously visited places (previous context information). This can be done individually for each location, the individual normality value listed in the table then being calculated by averaging from the individual normality values. For the normal day 220 and the largely normal day 240 results in a normality value of one, since all normally visited places were visited. On holiday 260 results also a value of one, since the places visited on this day were frequently visited in the past. For example, if the user visited the movie theater that day for the first time, for example, the normality value would have been 0.75 (four visited locations, three with a normality value of one and one with a normality value of zero).

In the column 112 find the normality values at times 112 that the places 111 assigned. During the day 220 and in the day 240 These are times that may include, for example, arrival times and / or departure times of each individual location, as usual. The result is a normality value of one. The value in line 240 could because of the slight deviations to the day 220 may also be slightly lower (eg 0.98) if, for example, more stages had been chosen in the discretization of the continuous information of the time. The times of the holiday day 260 are unusual, it gives a value of 0.2. For the times of the day 280 results in a normality value of 0.9. This may come as a surprise, as the day as a whole appears abnormal due to the unusual circumstances (mission to another location). The fact that a high value of 0.9 results is due to the exemplarily chosen statistical characteristic. In this example, conditional frequencies were chosen. For example, the current departure time in Restaurant B will be compared with those earlier departure times from that particular location (ie Restaurant B). Even though Restaurant B was seldom visited in the past (which is the value of 0 in the column 111 However, the departure time of Restaurant B may have always been similar.

In the column 113 find the normality values to the order 113 in which the places were visited. These normality values can be determined, for example, in the manner already described as a function of the Levenshtein distance. If the previous context information shows that a most frequent sequence is "HWRWH", then the results for the days 220 and 240 a Levenshtein distance of zero and hence a normality value of one. For the day 260 is the current context information "HRCH" (C for cinema, english cinema). The result is a Levenshtein distance of 0.5 and hence a normality value of 0.2. For the day 280 the current context information is eg "XYZYX". Compared with the previous context information, this results in a normality value of zero.

In the column 121 find the normality values for the travel modality 121 a way information 120 , Since the user usually (ie very often) covers half of all distances by car and public transport, the normality value for the day is 220 equal to one (routes 223 and 229 by car, routes 225 and 227 with public transport). During the day 240 and in the day 260 the user drives all routes by car, resulting in a normality value of 0.5. The normality value for the day 280 is again one. It may also be useful to consider other context information when determining the normality value for travel modality. For example, a statistical parameter may be formed from the travel modality at a particular start or destination. This results in conditional frequencies again, as previously explained.

In the column 122 find the normality values to the travel pattern 122 a way information 120 , These normality values can be calculated in a similar way as the normality values to the location order 113 , In the simplest case, the travel patterns are defined by the places lying between the trips. Then arise in the column 113 and 122 identical values. However, different context information can also be considered. For example, when calculating the values in column 113 only the places considered, so in column 122 the places and additionally the modality be considered. The statistical dependencies of individual normality values possibly resulting from the selected current and past context information may be desirable. They can also be taken into account by appropriate weighting in averaging.

In the column 123 you will find the normality values at the times 123 to which trips have been made. The calculation may be similar or even the same for the values in column 112 respectively. However, in the present example this is not the case. There were the values of the column 112 determined from conditional frequencies (departure time given a stay at a certain geographical location). On the other hand, the values of the column 123 necessarily determined. For this purpose, an individual information value was determined for all departure times of a day and a further individual information value for all arrival times of a day.

In the column 131 find the normality values at the event times 131 the event information 130 , In the present example, the only event time was 131 the getting up time 222 . 242 . 262 . 282 considered. As already described, however, numerous other events and their properties can be used as context information. It is also advantageous here, conditional frequencies as statistical To use characteristic. For example, the time of morning waking depending on the day of the week could be taken into account, since these times usually differ on weekdays and weekends.

The numerical example illustrates the calculation of the normality value of a whole day. It should be noted that the calculation of a normality value does not require this entire current context information of a whole day. Normality values can also be obtained from individual current context information, taking into account the similar information of the previous context information. This will be explained below with reference to another example from the daily routine 220 clarified.

In the first step of the method according to the invention, current context information is determined in this example. This includes an entry in the user's calendar that provides a meeting at the office for 11 o'clock. Furthermore, the current context information includes the alarm time 222 of the user.

In the second step, a previous context information concerning at least one past time is loaded. This includes the information on which days on which at least one meeting in the office is entered in the calendar, the user drives in the morning by car to the office. Further, the earlier context information includes all waking times of bygone days.

In the third step, a factual information is selected depending on the current context information and the previous context information. The factual information consists of a route recommendation for the route from home to the office.

In the fourth step, a normality value is determined as a function of the past context information and the current context information. In the present example, two individual normality values are determined for this purpose. The first value also refers to the alarm time. As explained above, this single normality value is equal to one. The second single normality value refers to the calendar entry (appointment at 11 o'clock). From the earlier context information, a statistical parameter is created for this purpose. As a result, on 97 out of 100 days, when at least one meeting in the office is entered in the calendar, the user drives to the office in the morning. It results from e.g. a single normality value of 0.97. The get-up time is weighted twice. This results in a normality value of 0.99.

In the fifth step, the factual information (recommended route) is not classified as relevant factual information because the normality value of 0.99 is above the threshold value of 0.5. In the final step, the factual information is not provided because it is not classified as relevant factual information.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• US 2013/0345961 A1 [0003]

Cited non-patent literature

• "Discovering Personal gazetteer: An Interactive Clustering Approach". C. Zhou et al, published in "Proceedings of the 12th Annual ACM International Workshop on Geographic Information Systems", pp 266-273, New York, NY, USA, 2004 [ 0019]
• AT Cluster et al., Published in "Proceedings of the 2008 ACM Symposium on Applied Computing," pp. 863-868, Fortaleza, Ceara, Brazil, 2008 [0019] "A Clustering-based Approach to Discovering Interesting Places in Trajectories"

Claims (10)

A method of providing factual information relevant to a user of a personal digital assistant comprising the steps of: - determining current context information concerning the present and / or a future time ( 100 ), - loading of past context information concerning at least one past time ( 100 ), - selecting a factual information in dependence of the current context information ( 100 ) and the previous context information ( 100 ), characterized by the steps of: determining a normality value as a function of the past context information ( 100 ) and the current context information ( 100 ), - decide on classification of the factual information as relevant factual information as a function of the normality value, and - provision of the factual information only when classification has taken place as relevant factual information. Method according to claim 1, characterized in that the current context information ( 100 ) and / or the previous context information ( 100 ) a place ( 110 . 111 . 112 . 113 ), a way ( 120 . 121 . 122 . 123 ) and / or an event ( 130 . 131 ) affect. Method according to one of the preceding claims, characterized in that the step of determining the normality value as a function of the past context information and the current context information comprises the following individual steps: determining a statistical parameter as a function of the previous context information ( 100 ) and determining the normality value as a function of the statistical parameter and the current context information ( 100 ). Method according to one of the preceding claims, wherein the current context information ( 100 ) at least two single context information ( 100 ), characterized in that the step of determining the normality value depends on the past context information ( 100 ) and the current context information ( 100 ) comprises the following individual steps: determining a single normality value for each of the at least two single context information ( 100 ) depending on the previous context information ( 100 ) and the current context information ( 100 ), - determining the normality value as a function of the at least two individual normality values. Method according to one of the preceding claims, characterized in that the normality value is a real number between zero and a constant upper limit and is preferably a number between zero and one. A method according to claim 5, characterized in that the step of deciding on a classification of the factual information as relevant factual information in dependence on the normality value comprises the following individual steps: - comparing the normality value with a constant threshold value and - grading the factual information as relevant factual information, if the normality value is smaller is considered the threshold or non-rating of the factual information as relevant factual information when the normality value is greater than the threshold. Method according to one of the preceding claims, characterized in that the step of providing the factual information includes a consideration of the normality value. Method according to one of the preceding claims, characterized in that the provision of the factual information is done visually, acoustically and / or haptically.  Personal digital assistant for carrying out the method according to one of claims 1 to 8.  A computer readable code storage medium for controlling a personal digital assistant to perform the method of any one of claims 1 to 8.

Images