METHOD AND DEVICE FOR INPUTTING A SEQUENCE OF SYMBOLS

Description

[0001] The present invention relates to a method for inputting a character string. In this method, a plurality of selectable characters are displayed on a touchscreen in input fields. Input positions on the touchscreen are then successively detected, and in an initial assignment, each input position is assigned to a displayed character, thus capturing an initial character string. Subsequently, a correction character string is determined. Furthermore, the invention relates to a device for inputting a character string using a touchscreen and an input unit coupled to the touchscreen. The input unit is configured to display a plurality of selectable characters on the touchscreen in input fields and to successively detect input positions on the touchscreen.The input unit is coupled to a detection unit configured to assign each input position to a displayed character in an initial mapping process to capture an initial character string. Furthermore, the device includes a correction unit configured to determine a correction character string.

[0002] To assist the user when entering a character string, it is known to compare the initial entered string with a set of valid strings. For example, if the string is a word from a specific language, the entered string can be compared, according to the initial assignment, with valid words of that language stored in a database. This method is used particularly in spell checking, where misspelled words are replaced with valid, correctly spelled words. The correctly spelled word then becomes the valid correction string, which is displayed as a suggestion. When the correction string is selected, the initially entered string is replaced by the correction string.

[0003] With such well-known methods for spell checking, the problem arises that in many cases correction strings are suggested that do not correspond to the user's input intention.

[0004] Document WO 2011 / 113057 A1 describes a method for inputting a character sequence. In this method, a user interface is tapped sequentially, and the input positions are recorded. A list of words is generated from these recorded input positions, corresponding to the inputs. This word list is intended to correct obvious user input errors. For this purpose, a character set is assigned to the input sequence, corresponding to the letters of a virtual keyboard, a set of approximation weights, and several feature types. The approximation weights represent the probability that a user intended to perform a different keystroke than the recorded keystroke. This probability depends on the proximity between the location of the input feature and the location of the letter on the virtual keyboard.

[0005] The invention is therefore based on the objective of providing a method and a device of the type mentioned at the outset, which determine which correction strings better correspond to the input intention of a user, if it has been determined that the initial string is not a valid string.

[0006] According to the invention, this problem is solved by a method with the features of claim 1 and a device with the features of claim 11. Advantageous embodiments and further developments are described in the dependent claims.

[0007] Accordingly, in the method according to the invention, a secondary assignment is performed when determining the correction string, in which at least one character of the primary assignment is replaced by another character according to the secondary assignment, so that a secondary string is generated. The selection of the other character according to the secondary assignment is carried out depending on the display position of the input field of the other character relative to the corresponding detected input position. The secondary string is then the correction string.

[0008] It was found that in many cases, the invalid initial string was not due to an orthographically incorrect spelling of a character string, but rather because the user had not selected the input field for entering the character they intended to select.

[0009] On many devices that use a touchscreen for character input, the available characters are displayed within input fields. The user can, for example, touch these input fields with their fingertip to select the character associated with the input field for the character sequence to be entered.

[0010] If the area on which the input fields are displayed is relatively small, as is the case with a smartphone, and the input fields are therefore also relatively small and arranged very close together, the problem arises that the user may not precisely hit the desired input position on the touchscreen and the character may be selected incorrectly. This problem also occurs when the device is used in an environment where acceleration forces are at work, such as vibrations, as occur when using it in a motor vehicle. In such cases of incorrect input, the method according to the invention can determine a correction string that better corresponds to the user's input intention.

[0011] In the inventive method, the detected input position is set in relation to the display positions of the input fields, and a different character is determined from this, which replaces an originally assigned character in the initial assignment. In this way, incorrect entries can be corrected that are not due to a lack of knowledge of the correct spelling of a character string, but rather result from the user, for example, typing next to the input field they actually wanted to select.

[0012] An input field is primarily an area on the touchscreen. However, it can also be a point on the touchscreen. An input field is characterized by the fact that it is associated with a character, which is selected when inputting text. This character is usually displayed in conjunction with the input field. For example, the character can be displayed within the area of ​​the input field on the touchscreen.

[0013] The input position refers specifically to the coordinates of the input on the two-dimensional touchscreen. First, an input area can be defined, which is, for example, the area where the user touches the touchscreen with their fingertip. From this input area, an input position is then determined. This could be, for example, the geometric center or the geometric centroid of the defined input area.

[0014] Alternatively, it is also possible that the user does not actually touch the touchscreen with their finger during input, but rather that an interaction between the surface of the touchscreen and the fingertip is detected and assigned to an input position. This interaction could, for example, be a capacitive coupling, which allows an input position to be detected even if the user only approaches the touchscreen with their fingertip but does not directly touch it. Such interactions for detecting an input position are also encompassed by a touchscreen within the meaning of the invention.

[0015] In the method according to the invention, the input fields are arranged in such a way that they form a keyboard in which the input fields are arranged very close to each other.

[0016] Advantageously, the method according to the invention makes it possible to better capture a user's input intention when an invalid character string has been detected because the user selected an incorrect input field, for example, because very small input fields are located very close to one another, or because the input was disturbed by forces that led to relative movements of the touchscreen to the input object, such as the user's fingertip. From the relative arrangement of the detected input position to the display positions of the input fields, the method according to the invention can determine a correction character string that better matches the user's input intention in these cases.

[0017] However, even if the initial string is a valid string, it may be highly improbable, so that in this case, too, the correction string is determined using the method according to the invention. For example, it may be a very rare word. Furthermore, it may be that this word is highly improbable in the context of the sentence. In this case, only after one or more secondary strings for the correction string have been determined can it be established which string best corresponds to the user's intention.

[0018] According to another embodiment of the method according to the invention, it is checked whether the initial string is a valid string. In this case, the correction string is only determined if the check shows that the initial string is not a valid string. Advantageously, this speeds up the method and reduces the computational effort.

[0019] According to a further development of the inventive method, a similarity measure is determined when comparing the first string with test strings from a set of permissible strings. Furthermore, in this case, a similarity measure is determined when comparing the second string with test strings from the set of permissible strings. One of the test strings is then selected as the correction string depending on the similarity measure, whereby the higher the similarity measure of the test string, the more likely it is to be selected as the correction string. Specifically, the test string with the highest similarity measure is selected. However, further parameters can optionally be considered when selecting the test string, as will be explained later.

[0020] The set of permissible strings could, for example, consist of words from a language stored in a thesaurus. In this case, the test strings are therefore the words of that language. Preferably, the first string is compared with the same test strings as the second string.

[0021] Established spelling correction methods can be used to determine the similarity score. A spelling correction is performed on the first string. A spelling correction is then performed on the second string, with a similarity score determined for each. The spelling correction string with the highest similarity score is then selected as the correction string.

[0022] Conventional spell checkers suffer from the problem that the position of a character within the character string is heavily weighted. For example, the initial letter is given relatively high priority, and often the suggested correction must not have been changed. Therefore, conventional spell checkers produce unsuitable correction strings if the initial letter has been incorrectly captured. In such cases, the suggested correction rarely matches the word the user intended to type. Similar weighting can occur depending on whether the letter is a vowel or a consonant.

[0023] In the aforementioned embodiment of the method according to the invention, information about the detected coordinates of the input position is taken into account. Depending on the relative arrangement of the input position to the display positions of the input fields, the secondary character string is determined. The same method, for example for spell checking, is then applied to this secondary character string to determine a similarity measure. However, if, for example, the initial letter of the secondary character string has been changed because the input position for entering the first character was relatively close to another input field at the edge of the input field of the first character of the primary character string, the character of this other input field will replace the initial letter of the primary character string in the secondary character string.This can advantageously lead to a situation where, due to the strong weighting of the initial letter in the correction procedure, a high degree of similarity is assigned to a check string that more closely corresponds to the user's input intention.

[0024] In this embodiment of the inventive method, not only is a linguistic comparison of character strings performed. Rather, the measurement of the input position and the geometric arrangement of input fields on the touchscreen are taken into account when generating the correction string. Considering these measurements leads to improved correction strings.

[0025] Alternatively, the similarity measure can also be determined in another way. For example, the procedure described in WO 2007 / 144199 A1 for automatically evaluating the similarity of two strings can be used to determine the similarity measure of the first string with each of the test strings and of the second string with each of the test strings.

[0026] In this embodiment of the inventive method, a character is replaced and it is checked whether a more similar test character sequence is obtained. If the second character sequence is identical to a test character sequence, the degree of similarity is maximized in this case, so that this test character sequence is then selected as the correction character sequence.

[0027] According to a further embodiment of the method according to the invention, it is checked whether the second string is a valid string, and the similarity measure is only determined if the check shows that the second string is not a valid string. Advantageously, this speeds up the method and reduces the computational effort.

[0028] According to a further embodiment of the method according to the invention, a confidence measure is assigned to the first string. When the second assignment of the second string is performed, a confidence measure for this second assignment is determined. The confidence measure of the second string is smaller than the confidence measure of the first string. Furthermore, the confidence measure of the second string is determined relative to the input position of the character that was assigned to the character that was replaced during the second assignment, relative to the position of the input field that was assigned to the character that was inserted during the second assignment.

[0029] The confidence level indicates how likely it is that the user intended to select the character in the respective string. During the initial assignment, it was assumed that the user intended to enter the character designated for the first string at the respective input position. Therefore, the first string has a higher confidence level than the second string, as otherwise the second string would be the first. However, the confidence level for the first string can also vary depending on the position of the input field. If the input field is always in the center of the input fields, the captured first string has a high confidence level. Conversely, if the input field was at the outer edge of at least some input fields, this leads to a decrease in the confidence level for the first string.

[0030] The confidence level for the second string is determined accordingly. Here too, the proximity of an input field to the input position of a character in the second string is taken into account. The closer the input field is to this input position, the higher the confidence level of the corresponding second string. Thus, if the input position of the input field selected for the first string is at the edge of that input field, adjacent to another input field, and the character of this other input field is used as a replacement in the second string, the confidence level of this second string will be lower than the confidence level of the first string. However, the reduction in confidence is not as large as when replacing a character assigned to an input field located on the opposite side of the edge of the input field used for the first assignment, or located further away from that input field.

[0031] The confidence measure of the first string is thus determined in particular depending on the input position which is assigned to the character in the first assignment, which was replaced in the second assignment, relative to the position of the input field of this character.

[0032] Advantageously, when determining the correction string, not only can a similarity measure between the first string or second string and the check strings be determined and taken into account, but a confidence measure can also be determined and taken into account, which is based on measurements of input positions and the determination of the relative display positions of input fields.

[0033] According to a further development of the inventive method, several second assignments are carried out, so that several second strings are generated, and a confidence measure is determined for each second string.

[0034] The selection of the check string as the correction string depends primarily on the similarity measure and the confidence measure. The higher the similarity measure of the check string, the more likely it is to be selected as the correction string. Furthermore, the higher the confidence measure of the first or second string, the more likely it is to be selected as the correction string. The confidence measure of the first or second string thus influences the selection of the check string. A confidence measure of the second string that is only slightly lower than that of the first string can therefore lead to the selection of a check string as the correction string if that check string has a higher similarity measure than another check string that has the highest similarity measure when compared to the first string.

[0035] In general, for a pair consisting of a first and second string with a check string, the similarity measure and confidence measure of the first and second strings can form the variables of a function whose function value determines whether the corresponding check string of this pair is selected as the correction string.

[0036] According to a further development of the inventive method, when determining the confidence measure, the direction of a vector from the center point of the input field of the character according to the first assignment to the input position is compared with the direction of a second vector from the center point of the input field according to the first assignment to the center point of the input field of the character according to the second assignment. The confidence measure of the second character sequence is then higher the less the direction of the second vector deviates from the direction of the first vector.

[0037] Alternatively or additionally, when determining the confidence measure, an initial distance of the input position from the center of the character's input field can be determined according to the secondary assignment. The confidence measure of the secondary character string is then higher the smaller this initial distance is.

[0038] Alternatively or additionally, when determining the confidence measure, a second distance between the input position and the nearest edge point of the input field of the character can be determined according to the second assignment. The confidence measure of the second character string is then higher the smaller the second distance.

[0039] In this way, the measured values ​​of the input positions can be related to the geometry of the input fields in order to determine the confidence measure and thus arrive at an improved selection of a correction string.

[0040] The input positions of the character string are temporarily stored so that they can be retrieved even after the initial assignment. Normally, the information about the input position is discarded as soon as a character is assigned to that position. In this case, however, the input positions can no longer be considered for selecting improved correction strings. Temporarily storing the input positions advantageously ensures that they remain accessible even after the initial assignment. This allows for the determination of secondary character strings based on these input positions, from which an improved selection of a correction string can then be generated.

[0041] According to a further development of the method according to the invention, the correction string is output as an acceptable proposal. If an input to accept the proposal is detected, the correction string is selected as the input string.

[0042] The device according to the invention, as described above, is characterized in that the correction unit is configured to perform a secondary assignment when determining the correction string, in which at least one character of the primary assignment is replaced by another character according to the secondary assignment, thus generating a secondary string. The selection of the other character according to the secondary assignment is carried out depending on the display position of the input field of the other character relative to the corresponding detected input position. The secondary string is then the correction string.

[0043] The device according to the invention is particularly designed to carry out the method according to the invention as described above. It therefore has the same advantages as the method according to the invention.

[0044] According to a further development of the device according to the invention, it has a test unit coupled to the detection unit, which is configured to check whether the initial string is a valid string. The correction unit is then coupled to the test unit and configured to determine the correction string only if the test unit's check shows that the initial string is not a valid string.

[0045] The correction unit of the device according to the invention is specifically configured to determine a similarity measure when comparing the first string with test strings from a set of permissible strings, and to determine a similarity measure when comparing the second string with test strings from the set of permissible strings. Furthermore, the correction unit is specifically configured to select one of the test strings as the correction string depending on the similarity measure, wherein the higher the similarity measure of the test string, the more likely this test string is to be selected as the correction string.

[0046] The similarity measure can, for example, only be determined if the test reveals that the secondary string is not a valid string. Conversely, the similarity measure can also be determined if the test reveals that the secondary string is a valid string.

[0047] Furthermore, the correction unit can be configured to determine a confidence measure, as explained above with reference to the method according to the invention.

[0048] The invention will now be explained using an exemplary embodiment with reference to the drawings. Figure 1 schematically shows the structure of an embodiment of the device according to the invention, Figure 2 shows the input device of the embodiment of the device according to the invention, Figure 3 shows the sequence of an embodiment of the method according to the invention, and Figure 4 illustrates the assignment of an input position to an input field, as is done in the embodiment of the method and device according to the invention.

[0049] With reference to the Figures 1 and 2 First, an exemplary embodiment of the device according to the invention will be explained: As in Figure 1 As shown, the device according to the invention comprises a touchscreen 1. The touchscreen 1 comprises a display area and a touch-sensitive surface arranged on the display area.

[0050] The touchscreen 1 is connected to a display controller 2 and an input unit 3. The display controller 2 controls the display on the display area of ​​the touchscreen 1. The input unit 3 is also connected to the display controller 2. It can cause the display controller 2 to display responsive input fields 10 in a specific area of ​​the display area of ​​the touchscreen 1, in which selectable characters are displayed. The user can enter data via these input fields 10, i.e., select characters, which are then displayed by the display controller 2 in a display area 9 of the touchscreen 1, as shown in Figure 2 shown.

[0051] Using the touchscreen 1, a user input can be assigned an input position on the touchscreen. For example, the user can touch the surface of the touchscreen 1 with their fingertip in a touch area 11, as shown in Figure 2as shown. However, physical contact is not strictly necessary. It may be sufficient for the user to bring their fingertip or another object near the touchscreen 1, so that an electromagnetic coupling, for example a capacitive coupling, is generated, by which the position of the fingertip or the object can be detected. Such touchscreens 1 are known per se.

[0052] The input unit 3 is designed to detect the input area 11 and assign it to an input position. These are, in particular, the xy coordinates on the surface of the touchscreen 1.

[0053] The input unit 3 is coupled to a detection unit 4. The detection unit 4 is configured to assign each input position, in conjunction with the current display on the touchscreen 1, to a specific input field 10 during an initial assignment. Each input field 10 is assigned a character, which is displayed, for example, within the input field 10. By successively detecting characters and performing the initial assignment, the detection unit 4 captures an initial character string.

[0054] The acquisition unit 4 is coupled with the display control 2. The display control 2 can display the characters acquired by the acquisition unit 4 in the display area 9 of the touchscreen 1.

[0055] The acquisition unit 4 is also coupled to a verification unit 5. The verification unit 5 is configured to check whether the initial string transmitted by the acquisition unit 4 is a valid string. For this purpose, the verification unit 5 is coupled to a data storage device 6. The data storage device 6 contains a set of valid strings, for example, all words of a specific language.

[0056] The test unit 5 is coupled to the display control 2. If the test unit 5 has determined that the initial string is a valid string, the test unit 5 can transmit a corresponding signal and the initial string to the display control 2, so that the initial string is displayed in the display area 9 by means of the display control 2.

[0057] Furthermore, the test unit 5 is coupled to a correction unit 7. The correction unit 7 is configured to determine a valid correction string if the test unit 5 determines that the initial string is not valid. As will be explained later, the correction unit 7 generates a valid correction string. The correction unit 7 is also coupled to the display control 2, allowing it to transmit the correction string to the display control 2. The display control 2 can then output the correction string as an acceptable suggestion.

[0058] The touchscreen 1, the display control 2, the input unit 3, and the acquisition unit 4 are integrated, in particular, into an input device 8, for example, a smartphone. Furthermore, the testing unit 5, the data storage 6, and the correction unit 7 can also be integrated into the input device 8. However, these units can also be located outside the input device 8, in which case a data interface is provided between the input device 8 and the externally located units. Alternatively, only the data storage 6 can be located outside the input device 8.

[0059] As in Figure 2As shown, the input device 8 generates a display area 9 on the display surface of the touchscreen 1, in which the captured initial strings can be displayed. Furthermore, the correction string can also be displayed in the display area 9 as a suggestion for a replacement for an initial string. Alternatively, several correction strings for an initial string can also be displayed in the display area 9.

[0060] Furthermore, a keyboard with input fields 10 is displayed on the screen of the touchscreen 1. To enter a character from the input fields 10, the user can touch the touchscreen in the area of ​​the keyboard at the touch area 11. This touch area 11 is then assigned an input field 10 and a character associated with that input field 10, as will be explained later.

[0061] With reference to the Figures 3 and 4An embodiment of the method according to the invention will now be explained, with further details of the device according to the invention being discussed. The device according to the invention is specifically designed to carry out the method according to the invention.

[0062] In step S1, a keyboard with a variety of selectable characters is displayed on the touchscreen using the input unit 3 and display control 2 in input fields 10. The character assigned to an input field 10 is displayed within that input field, allowing the user to select the corresponding character by touching the touchscreen 1 in the area of ​​the relevant input field 10. The user then touches the touch-sensitive surface of the touchscreen 1 in a touch area 11, for example, with their fingertip.

[0063] In step S2, the area of ​​this touch area 11 on the touch-sensitive surface of the touchscreen 1 is detected by means of the input unit 3.

[0064] In step S3, this flat contact area 11 becomes an input position 12 (see Figure 4 ) calculated. In the present embodiment, the input position 12 is the geometric center or the geometric centroid of the area of ​​contact 11. The input position 12 is temporarily stored.

[0065] In step S4, the captured input position 12 is assigned an input field 10, and thus a displayed character, by means of the capture unit 4 in an initial assignment. Furthermore, after a character has been assigned to an input position 12, this character is output in the display area 9, so that the user can immediately see each entered character in the display area 9.

[0066] Steps S2 to S4 are repeated until the selected character indicates that a complete string has been entered. For example, if a space was selected, it follows that the previously selected characters should represent a complete word. In this case, the process continues with step S5, in which an initial string is generated from the first mapping of the captured characters.

[0067] In step S6, the initial string is then checked to determine if it is a valid string. This is done by comparing the initial string with check strings. These check strings are read from data memory 6 and are, for example, the valid words of a language. If step S6 determines that the initial string is valid, it is transferred to the display controller 2, which in turn instructs the input device 8 to output the initial string as a complete string on the touchscreen 1 in the display area 9 in step S7.

[0068] The process then continues with step S2, allowing a new string to be entered.

[0069] If step S6 reveals that the initial string is not a valid string, the procedure continues with step S8.

[0070] In an alternative embodiment, step S6 is omitted, and the process continues with step S8 even if the initial string is a valid string. Regardless, the string is still output in step S7.

[0071] In step S8 and the following steps, a valid correction string is determined. A secondary mapping is performed in which at least one character of the primary mapping is replaced by another character according to the secondary mapping, thus generating a secondary string. To explain how the secondary mapping is performed, the following section provides further information. Figure 4 Reference made to.

[0072] In Figure 4Various input fields 10 are shown, each assigned a selectable character, in the illustrated embodiment a letter. This letter is displayed within the input field 10. The touch position 12 determined from the touch area 11 is, in the illustrated example, located within the control field 10 for the letter G. During the initial assignment, the letter G is selected according to the touch area 11. As shown in Figure 4 However, as shown, the touch position 12 is not in the middle of the input field 10 for the letter G, but near the edge.

[0073] In step S8, the geometry of the input fields 10, as displayed on the touchscreen 1, is determined relative to the touch position 12. During the second assignment, the letter G of the first assignment is replaced by another character, with the selection of this other character being carried out according to the second assignment depending on the display position of the other character in input field 10 relative to the corresponding detected input position 12.

[0074] In step S9, the direction of a first vector from the midpoint 14 of input field 10 of character G (according to the initial assignment to input position 12) is compared with the direction of a second vector from the midpoint 14 of input field 10 (according to the initial assignment) to the midpoint of the input field of character G (according to the second assignment). A confidence measure parameter is determined from the comparison of the directions of these two vectors. The confidence measure indicates the probability that the user intended to select the other character.

[0075] In the example according to Figure 4A confidence measure can thus be determined that the other character is the letter H or the letter F. It turns out that the two previously described vectors for input field 10 of the letter H exhibit a smaller directional deviation than the corresponding vectors belonging to input field 10 of the letter F. Specifically, the point of contact 12, relative to the center point 14 of input field 10 for the letter G, lies more in the direction of input field 10 for the letter H than in the direction of input field 10 for the letter F. Accordingly, the letter H, as the character that replaces the letter G of the first assignment according to the secondary assignment, is assigned a higher confidence measure than the letter F. The confidence measure of the secondary character sequence is generally higher the less the direction of the second vector deviates from the direction of the first vector.

[0076] Another parameter used to determine the confidence measure is the initial distance of input position 12 from the center of input field 10 of the character, according to the second assignment. As shown in Figure 4 As can be seen, this parameter also leads to a higher confidence level for the letter H than for the letter F. In general, the confidence level of the second string is higher the smaller the first interval.

[0077] Another parameter used to determine the confidence measure is the second distance between input position 12 and the nearest edge point 15 of input field 10 of the character, according to the second assignment. In this case, too, this leads to a higher confidence measure for the letter H compared to the confidence measure for the letter F, since input position 12 is closer to the nearest edge point 15 of input field 10 for the letter H than to the nearest edge point of input field 10 for the letter F. The smaller this second distance, the higher the confidence measure of the second character string.

[0078] In this way, step S9 performs a confidence measure for the input of each character in the vicinity of a character according to the initial mapping. Step S9 is then repeated for all characters, with a confidence measure being determined for characters in neighboring fields each time.

[0079] Subsequently, in step S10, a confidence measure is determined for each character of the initial assignment for the first string. The confidence measure of the characters in the first string is determined relative to the input position 12 assigned to the character in the initial assignment, relative to the position of that character in input field 10. The further the input position 12 is from the center point 14 of input field 10 according to the initial assignment, the lower the confidence measure of that character according to the initial assignment. However, the confidence measure of the character according to the initial assignment is always higher than the confidence measure of another character according to the secondary assignment, which replaces this character in the secondary assignment.

[0080] In the example according to Figure 4The confidence level for the character G is thus reduced because the touch position 12 is not directly in the center of the input field 10 for the character G, but rather towards the edge of the input field 10 for the letter G. However, the confidence level for the character H is lower than the confidence level for the character G according to the initial assignment, since otherwise the character H would have been selected during the initial assignment. With the completion of step S10, the confidence level for each character of the initial character sequence is therefore also available.

[0081] In step S11, a secondary mapping is performed based on the determined confidence measures. A character from the primary mapping is replaced by another character, specifically the character whose confidence measure differs least from that of the character in the primary string it would replace. The confidence measure of the secondary string is then derived from the confidence measures of the individual characters within that secondary string. Similarly, in step S11, another secondary string or several further secondary strings are determined, with each secondary string assigned a confidence measure that depends on the confidence measures of the replaced characters.

[0082] In step S12, it is then checked whether the secondary string or one of the secondary strings is a valid string. If so, step S13 is continued, and the valid secondary string becomes the correction string. This is transferred to display control 2, which outputs it in display area 9 as an acceptable suggestion for replacing the primary string.

[0083] If the second string or none of the second strings is a valid string, proceed to step S14.

[0084] In an alternative embodiment, steps S12 and S13 are omitted, and the process continues with step S14 even if the secondary string or one of the secondary strings is a valid string.

[0085] In step S14, a similarity comparison is performed for the first string and the second string(s) against check strings stored in data memory 6 as valid strings. This similarity check can be performed for all strings stored in data memory 6 or only for a subset of these strings. For each pair of the first string with a check string, a similarity measure is determined for that check string. Similarly, for each pair of a second string with a check string, a similarity measure is determined. The similarity measure can be determined in the same way as in conventional spell checking. Alternatively or additionally, the similarity measure can be determined using a procedure as described in WO 2007 / 144199 A1.

[0086] In step S15, a function is applied to each test string, taking as variables the similarity measure of that test string and the confidence measure of the first and second strings with which the corresponding test string was compared. The higher the similarity measure and the higher the corresponding confidence measure, the higher the value of this function. The test string with the highest function value is then selected as the correction string in step S15.

[0087] The procedure then continues with step S13, in which the correction string is transferred to the display control 2, which controls the touchscreen 1 to output the correction string in display area 9 as an acceptable suggestion for a replacement of the initial string.

[0088] Starting from step S13, the process continues with step S16. In step S16, any input for accepting the generated proposal is recorded. If input for accepting the proposal has been recorded, in step S17 the initial string is replaced by the correction string and displayed accordingly in display area 9.

[0089] In another embodiment, the test string is determined solely based on the similarity measure, and the steps relating to the confidence measure are omitted. The test string with the highest similarity measure is then selected as the correction string.

[0090] For further explanation, reference is made again to the section in Figure 4 The example shown explains how the correction string is determined in the inventive method and device.

[0091] It is assumed in the following that the user intended to enter the following string as the target string: HALF

[0092] However, for the first letter H, the user did not touch input field 10 for the letter H, but rather touch area 11 and thus touch position 12, as shown in Figure 4 This is shown in input field 10 for the letter G. The following initial character string was thus captured: GALB

[0093] The confidence level for this initial string is 0.9, since input position 12 is not in the center of input field 10 for the letter G. It is assumed that for the subsequent letters ALB, input positions 12 were each in the center of the corresponding input fields for those letters.

[0094] The procedure now determines that the initial string GALB is not a valid string. Therefore, two different strings are identified, their respective confidence measures are calculated, and the initial string GALB and the identified two strings are compared with several test strings stored as valid strings in data memory 6.

[0095] The following table shows the similarity measure of the initial string GALB with various check strings: Test mark sequences Similarity measure HALF 0,6 (one character different, first character different) YELLOW 0,8 (one sign different) CALF 0,6 (one character different, first character different) LEAVES 0,4 (two characters different, first character different) GUEST 0,45 (two characters different)

[0096] Furthermore, as explained above, several secondary strings are determined. The first secondary string is: HALF

[0097] As mentioned above with reference to Figure 4As explained previously, in this case the letter G in the first string has been replaced by the letter H. The confidence level of this second string is 0.8. It is lower than the confidence level for the first string. This second string is then compared to the same test strings as the first string. A similarity measure is determined for each pair of second strings with a test string. This is shown in the following table: Test mark sequences Similarity measure HALF 1 (identical) YELLOW 0,4 (two characters different, first character different) CALF 0,6 (one character different, first character different) LEAVES 0,4 (two characters different, first character different) GUEST 0,3 (three characters different)

[0098] The test string HALB has a similarity score of 1, since this test string is identical to the secondary string. In the inventive method, this would already result in this secondary string being output as the correction string.

[0099] However, even if the similarity measure were not 1 due to the string's identity, but somewhat lower than this maximum similarity measure, the value of the function described above for the test string HALF would be higher overall than the value of the function for the test string YELLOW, which has the highest similarity measure when compared to the first string. For this reason, in this case too, the test string HALF will be output as the suggested correction instead of the test string YELLOW.

[0100] Another possible second string could be, as in Figure 4 As shown, the letter F replaces the letter G, since it is also located next to control panel 10 for the letter G. The secondary character sequence in this case would be the following: FALB

[0101] The confidence level for this secondary string would be, for example, 0.6. It is lower than the confidence level for the secondary string HALF, because input field 10 for the letter F is further away from touch position 12 than input field 10 for the letter H.

[0102] In this case, too, the similarity measure for this secondary string is determined using test strings. The result is shown in the following table: Test string n Similarity measure HALF 0,6 (one character different, first character different) YELLOW 0,4 (two characters different, first character different) CALF 0,6 (one character different, first character different) LEAVES 0,4 (two characters different, first character different)

[0103] As a result, the function described above with the variables of the similarity measure and the confidence measure will not yield a higher value, so that even when this additional second string is taken into account, HALF will be output as the correction string. Reference symbol list

[0104] 1 Touchscreen 2 Display control 3 Input unit 4 Capture unit 5 Verification unit 6 Data storage 7 Correction unit 8 Input device 9 Display area 10 Input fields 11 Touch area 12 Touch position 13 Central area 14 Center of an input field 15 Edge of an input field

Claims

1. Method for inputting a character sequence, in which a plurality of selectable characters are displayed on a touch screen (1) with input fields (10), input positions (12) are recorded one after other on the touch screen (1) and, in a first allocation, each input position (12) is allocated to a displayed character, such that a first character sequence is recorded, a confidence measure is allocated to the first sequence, and for the first character sequence, a correction character sequence is ascertained, wherein, upon ascertaining the correction character sequence, a second allocation is carried out in which at least one character of the first allocation is replaced by a different character according to the second allocation, such that a second character sequence is generated, wherein the selection of the other character according to the second allocation is carried out depending on the display position of the input field (10) of the other character relative to the corresponding recorded input position (12), wherein, when carrying out the second allocation of the second character sequence, a confidence measure of this second allocation is ascertained, wherein the confidence measure of the second character sequence is smaller than the confidence measure of the first character sequence, and wherein the confidence measure of the second character sequence is determined depending on the input position (12), which is allocated to the character in the first allocation, which has been replaced in the second allocation, relative to the position of the input field, which is allocated to the character, which has been used in the second allocation, and the second character sequence is the correction character sequence, characterised in that when ascertaining the confidence measure, the direction of a first vector from the centre point of the input field (14) of the character is compared according to the first allocation in relation to the input position (12) to the direction of a second vector from the centre point of the input field (14) according to the first allocation in relation to the centre point of the input field (14) of the character according to the second allocation, and the higher the confidence of the measure of the second character sequence, the less the direction of the second vector deviates from the direction of the first vector, or upon ascertaining the confidence measure, a second distance of the input position (12) from the closest edge point of the input field (15) of the character is determined according to the second allocation, and the higher the confidence measure of the second character sequence, the smaller the second distance.

2. Method according to claim 1, wherein it is checked whether the first character sequence is a permissible character sequence, and the correction character sequence is only ascertained when the check reveals that the first character sequence is not a permissible character sequence.

3. Method according to claim 2, wherein, in a comparison of the first character sequence with check character sequences from an amount of permissible character sequences, in each case a similarity measure is ascertained and, in a comparison of the second character sequence with check character sequences from the amount of permissible character sequences, in each case a similarity measure is ascertained, and a selection of one of the check character sequences as the correction character sequence is carried out depending on the similarity measure, wherein the higher the similarity measure of the check character sequence, the sooner this check character sequence is selected as the correction character sequence.

4. Method according to claim 3, wherein it is checked whether the second character sequence is a permissible character sequence, and the similarity measure is only ascertained when the check reveals that the second character sequence is not a permissible character sequence.

5. Method according to one of the preceding claims, wherein the confidence measure of the first character sequence is determined depending on the input position (12), which has been allocated to the character in the first allocation, said character having been replaced in the second allocation, relative to the position of the input field (10) of this character.

6. Method according to one of the preceding claims, wherein several second allocations are carried out, such that several second character sequences are generated, and for each second character sequence, a confidence measure is respectively ascertained.

7. Method according to one of claims 3 to 6, wherein the selection of the check character sequence as the correction character sequence is carried out depending on the similarity measure and on the confidence measure, wherein the higher the similarity measure of the check character sequence, the sooner this check character sequence is selected as the correction character sequence, and the higher the confidence measure of the first or second character sequence, with which the check character sequence has been compared, the sooner this check character sequence is selected as the correction character sequence.

8. Method according to one of claims 1 to 7, wherein, when ascertaining the confidence measure, a first distance of the input position (12) from the centre point of the input field (10) of the character is determined according to the second allocation, the confidence measure of the second character sequence is higher the smaller the first distance is.

9. Method according to one of the preceding claims, wherein the input positions (12) of the character of the character sequence are temporarily stored so that they can still be retrieved even after the first allocation.

10. Method according to one of the preceding claims, wherein the correction character sequence is issued as a reasonable suggestion and, if an input is recorded to accept the suggestion, the correction character sequence is selected as the input character sequence.

11. Device for inputting a character sequence having: a touch screen (1), an input unit (3) coupled to the touch screen, which is formed to display a plurality of selectable characters on the touch screen (1) with input fields (10) and to record input positions (12) on the touch screen (1) one after the other, a recording unit (4) coupled to the input unit (3), which is formed to allocate each input position (12) to a displayed character in a first allocation for recording a first character sequence, and a correction unit (7) which is formed to ascertain a correction character sequence, wherein the correction unit (7) is formed to allocate a confidence measure to the first character sequence and to carry out a second allocation upon ascertaining the correction character sequence, in which at least one character of the first allocation is replaced by a different character according to the second allocation, such that a second character sequence is generated, wherein the selection of the other character is carried out according to the second allocation depending on the display position of the input field (10) of the other character relative to the corresponding recorded input position (12), wherein, when carrying out the second allocation of the second character sequence, a confidence measure of this second allocation is ascertained, wherein the confidence measure of the second character sequence is less than the confidence measure of the first character sequence, and wherein the confidence measure of the second character sequence is determined depending on the input position (12), which is allocated to the character in the first allocation, which has been replaced in the second allocation, relative to the position of the input field, which is allocated to the character, which has been used in the second allocation, and the second character sequence is the correct character sequence, characterised in that, when calculating the confidence measure, the direction of a first vector from the centre point of the input field (14) of the character according to the first allocation to the input position (12) is compared to the direction of a second vector from the centre point of the input field (14) according to the first allocation to the centre point of the input field (14) of the character according to the second allocation, and the higher the confidence measure of the second character sequence, the less the direction of the second vector deviates from the direction of the direction of the first vector, or when ascertaining the confidence measure, a second distance of the input position (12) from closest edge point of the input field (15) of the character is determined according to the second sequence allocation, and the higher the confidence measure of the second character sequence, the smaller the second distance is.

12. Device according to claim 11, wherein the device has a check unit (5) coupled to the recording unit (4), which is formed to check whether the first character sequence is a permissible character sequence, and the correction unit (7) is coupled to the check unit (5) and is formed to only ascertain the correction character sequence when the check of the check unit (5) reveals that the first character sequence is not a permissible character sequence.

13. Device according to claim 11 or 12, wherein the correction unit (7) is formed to respectively ascertain a similarity measure when comparing the first character sequence to the check character sequence from an amount of permissible character sequences, and to respectively ascertain a similarity measure when comparing the second character sequence to check character sequences from the amount of permissible character sequences, and to make a selection of one of the check character sequences as the correction character sequence depending on the similarity measure, wherein the higher the similarity measure of the check character sequence, the sooner this check character sequence is selected as the correction character sequence.