JP2002373049A - Character string input device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To miniaturize a keyboard part and to facilitate the converting operation in character string input including Japanese KANA/KANJI (syllabary/ Chinese character) conversion as well. SOLUTION: When a plurality of characters are assigned to one key (figure 1-2), a plurality of switches or the like (figure 1-1) are assigned to that key, the probability to select each of characters assigned to the key is found by estimating a place, where the key is pressed, from the states of a plurality of switches (figure 1-4), and a character string is estimated (figure 1-6) by using a language limitation and a language probability (figure 1-5).

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to character string input including Japanese Kana-Kanji conversion, and more particularly to miniaturization of a keyboard portion and simplification of conversion operation.

[0002]

2. Description of the Related Art Web page http: // www. big. or. jp / ^~ cra
ne / diary / 1999 / adf_shkey. h
A part of the description of the conventional character string input device described in tml is cited in 0003 below, and an image is shown in FIG. FIG. 2 shows a configuration diagram of this conventional character string input device. In the conventional character string input device, a key unit (FIG. 2-7) forms a keyboard unit.

Special key: Ar is a kana-kanji conversion key. UpC is backspace and conversion cancellation. Tc switches input modes. Sp performs space and conversion confirmation. The problem here is why the key imprint is such a meaningless one.
It is very confusing. Even if you touch it with interest at the store, you will not know how to use it with this stamp. I didn't even understand (＾ _ ＾ ;. This alone gives me a difficult impression. I should aim for something that can be used by touching it. Input method: There are few keys and more than one key It's a little strange because there is a key with a letter written on it, but in fact this is an interesting trick, for example, a key with the letter "ZK". When you press this, the letter "Z" appears on the screen. However, it is an ambiguous key to enter either “Z” or “K.” For example, “KYOUSE
I) ". The key is "ZK, YM, S
O, UD, SO, E, HI. Look up a word in the dictionary that matches this Roman character. Then, "ZYO-U-SE-I"
"ZYO-UOEI" (Jyoei) "ZYO
-DO-EI (Jodoei) "," KYO-U-SE-
I (Kyosei: compulsory, correction _‥ ) ”,“ KYO-UO-
EI (Kyoei) and "KYO-DO-EI". As a result, it will be "Judge" or "Judge", and these will be displayed as conversion candidates.

In the conventional example shown in FIG. 2, a keyboard section (FIG. 2-3) composed of a plurality of key sections (FIG. 2-7) to which one or more characters are assigned, and language information such as language restrictions. And a character string estimating unit (FIG. 2-6) for estimating an input character string.

In the above-mentioned conventional example, one or more characters are assigned to one key portion. However, since there is no information as to which character of the pressed key portion has been input, the character is input from the keyboard portion. The character string estimating unit performs the determination of the character string and the kana-kanji conversion using the word constraint information and the language constraint information of the language model unit.

[0006]

However, in the above-described conventional character string input device, since it is not known which of the characters assigned to the pressed key portion has been input, the input is not possible. Only language constraints can be used for estimating the estimated character string, and the processing for estimating the character string is heavy, and the estimation accuracy is low because a large number of estimated character string candidates are generated.

Accordingly, the present invention increases the processing speed for character string estimation and provides more accurate character string estimation and conversion.

[0008]

According to a first aspect of the present invention, there is provided a character string input device, comprising: information on which part of each sensor key is pressed, and at what angle the sensor key is pressed; Based on, the probability estimating unit assigns the specified probability to each character assigned to the sensor key unit, so that in addition to the linguistic constraints and linguistic probabilities, the specified probability of the character is By using this, it is possible to estimate a character string with high accuracy with a small amount of calculation. The sensor key unit and the keyboard unit according to the second and third aspects make it possible to reduce the burden on the user of the present invention.

[0009]

BEST MODE FOR CARRYING OUT THE INVENTION

FIG. 1 shows a configuration diagram of an embodiment of the present invention.

In the character string input device according to the first aspect, the switch unit (FIG. 1-1) which is a component of the sensor key unit (FIG. 1-2) arranged on the keyboard unit (FIG. 1-3) includes:
The information at the time when the sensor key is pressed is used as a probability estimator (FIG. 1).
-4). The probability estimating unit obtains, from information received from each sensor key unit, the probability of inputting that character for all the characters assigned to that sensor key unit. The character string estimating unit (FIG. 1-6) receives information from the probability estimating unit and the language model unit (FIG. 1-5), and stores the probability for each character assigned to the sensor key unit and the language model unit. The input character string is estimated using the probability of a certain statistical language model, contexts before and after, word constraints, linguistic knowledge, language constraints, and the like.

As shown in FIG. 3, the sensor key portion according to the first embodiment is composed of three assigned characters "A", "B", and "C" and two switches on the left and right (FIG. 3). -8, FIG. 3-9), the operation of the sensor key unit will be described as an example. When the user presses the left side of the sensor key unit to input "A", the switch in FIG. 3-8 will be pressed. At this time, the sensor key unit passes information that the switch in FIG. 3-8 is pressed to the probability estimating unit. In addition, after the switch of FIG. 3-8 is pressed, although the left side of the sensor key portion is pressed, FIG.
The switch 9 may be pressed. In this case, the sensor key unit passes information on the time when both switches are pressed and released to the probability estimating unit.

Assume that the user presses the center of the sensor key to input "B". Then, FIG. 3-8 and FIG.
Both switches -9 are pressed. Also in this case, the sensor key unit passes information on the time when both switches are pressed and released, to the probability estimating unit. Also, even if the center of the sensor key is pressed, only one of the switches may be pressed, but in this case, the sensor key only outputs the information that the pressed switch has been pressed by the probability estimating unit. Pass to.

Next, a case where two pressure sensors (FIGS. 3-8 and 3-9) are provided in the sensor key portion will be described.
In this case, exactly the same method as the switch, that is, when one of the sensors senses the pressure, passes it to the establishment estimator, and when both sensors sense the pressure, passes the time information to the establishment estimator. Used. In addition, in the case of a pressure sensor, pressure information can also be passed to the probability estimator to the probability estimator.

Which position on the sensor key is pressed,
Means for obtaining information such as the angle at which the sensor key is pressed is not limited to a plurality of switches and pressure sensors, and any means may be used. For example, any number of angle sensors, switches, sensors, and the like that can detect the pressed angle can be used.

FIG. 3 shows two switches or pressure sensors.
Although the sensor key portion in the case of three assigned characters is shown, the number of switches or pressure sensors and the assigned characters are not limited. In FIG. 3, the assigned characters are arranged in one line, but need not be arranged in one line. For example, when four characters are assigned to the sensor key portion, the shape of the sensor key portion may be any shape such as a square or a circle.

A probability estimating unit according to claim 1 (FIG. 1-4).
The following describes an example of a method of calculating the input probability of each character assigned to the sensor key unit from information received from the sensor key unit.

In the method of determining the probability of each character assigned to the sensor key portion being pressed, the establishment estimating portion estimates which portion of the sensor key portion has been pressed, what angle the sensor key portion has been pressed, and the like. It is done by doing.
The operation of the establishment estimation unit (FIGS. 1-4) when two switches are used for the sensor key unit will be described with reference to FIG. When receiving the information from the sensor key unit, the probability estimating unit determines the probability for each character assigned to the pressed sensor key unit.

If the center of the sensor key portion shown in FIG. 3 cannot be pressed, only the left or right switch (FIGS. 3-8 and 3-9) will be pressed. At this time, for example, when the left switch is pressed, "A"
70% and "B" a 30% probability, and a character with low use frequency is assigned to the center, etc. Except for the above, only the case where there are three ways of pressing, a left switch, a right switch, and both left and right switches, will be described. When only the switch in FIG. 3-8 is pressed, "A" is set to 1
A probability of 00%, a probability of 0% for "B", and a probability of 0% for "C" can be given, and 80% for "A" and 10% for "B".
% And “C” can be given a probability of 10%. It is preferable that the degree of the provision of the probability can be operated by the probability estimating unit and can be changed by the user.

When only the switch shown in FIG. 3-9 is depressed, the left and right may be reversed in the above description, or an asymmetric probability may be assigned.

The method of assigning the probability assigned to each character, which is estimated by the probability estimating unit, may be different for each sensor key unit or may be the same.

Using the time information from when the switch or the like assigned to the sensor key section is pressed and released, and the time sequence information when the switch or the like assigned to the sensor key is pressed, the user can really switch each switch. You can guess whether you wanted to press. For example, even when both switches are pressed, if one of the switches has been pressed for a very short time, the user can consider that he did not want to press that switch.

Assume that the switch in FIG. 3-8 is pressed first, and then or simultaneously with the switch in FIG. 3-9. After this, you have two options: 1: The switch of FIG. 3-9 is turned off first, and then the switch of FIG. 3-8 is turned off.
2: The switch in FIG. 3-8 is turned off first, and then the switch in FIG. 3-9 is turned off. In the case of 1: the characters having a high probability of being input are "A" and "B". The height of the input probability can be considered to be, for example, “A” ≧ “B”. At this time, for example, “A” is 60%, “B” is 20%,
C "can be given a probability of 20%. 2:
In the case of, the characters having a high probability of being input are "A" and "B".
It is. The input probability is, for example, "A" and "
B "can be considered to be comparable. At this time, for example, 40% for" A ", 40% for" B ", 20% for" C ", or 30% for" A ", and" B " 50%, "C"
Can be given a probability of 20%. The strength of these probability assignments is arbitrary and can depend not only on the time order but also on the length of time the switch is pressed. It is preferable that the user can change the way of giving the time-dependent probability and the degree of the strength.

Assume that the switch in FIG. 3-9 is pressed first, and then or simultaneously with the switch in FIG. 3-8. In this case, in the above description, the left and right may be reversed, a left-right asymmetric processing method and a probability assignment method may be used.

When a pressure sensor is used for the sensor key portion, sensitivity information of the pressure sensor can be used in addition to the same processing as when a switch is used. Weighting may be performed so that the higher pressure is closer to the pressed position and the probability of the character assigned to that position is higher. In this case, it is possible to estimate the pressed position more accurately than using the switch.

When one pressure sensor capable of detecting pressures at a plurality of locations is used for the sensor key section (FIG. 4), the same processing as described above may be performed. Also, the same processing method as described above can be used when switches or sensors that can detect an angle or the like when the sensor key portion is pressed are used.

In the above description, an example was described in which a rectangular sensor key portion to which three characters were assigned was used. However, the number of characters assigned to one sensor key portion and the shape of the sensor key portion are arbitrary. For example, the shape of the sensor key portion to which four characters are assigned includes a circle and a square. Further, the shape of the sensor key portion does not need to be symmetric, and may be asymmetric.

After the user inputs the character string, the character string estimating unit receives the linguistic constraint information from the language model unit (FIG. 1-5), and the character string of the combination of characters assigned to the sensor key unit input by the user. From among them, all possible word strings including one or more words or all word strings having a probability equal to or higher than a threshold are stored together with probability information. After that, the character string estimation unit receives linguistic knowledge such as statistical language model, empirical linguistic knowledge, and linguistic word connection constraints from the language model unit, and by using the context of the surrounding sentence,
It estimates which word string is the word string that the user wants to input, and notifies the user in order of candidate word strings with high probability.

The processing example in the case where three alphabets are assigned to one sensor key portion has already been described. Here, the processing example in the case where five kana are assigned to one sensor key for inputting Japanese kana. I do.

"A" and "I" are displayed on the right side of the sensor key part.
FIG. 21 shows an example in which “U” is assigned to the center and “E” and “O” are assigned to the left. First, an example will be described in which the left side of the sensor key unit is accurately pressed, and at that time, the probability estimation unit determines that either the left side or the center is pressed. As in the case of the conventional mobile phone, pressing the left side of the sensor key portion once is likely to select “A”, and pressing the left side again is likely to select “I”. Here, when the left side is pressed only once, there is a possibility that “A” has been pressed and there is a possibility that “U” has been pressed, but when the left side is pressed twice, the center of the sensor key portion is displayed. Since two pseudonyms are not allocated, it is possible to eliminate the possibility that the user wanted to press the center. In other words, the user can accurately set the left side of the sensor key
When the button is pressed twice, the probability of pressing "i" can be set to 100%. The above processing method is an example, and the user wants to press “U” and presses the center of the sensor key portion twice,
Of course, it is possible. Next, an example of processing when both the left and right switches are pressed will be described. Once both left and right switches are pressed, for example, "A", "I",
"E" and "O" can be assigned a probability of 10%, and "U" can be assigned a probability of 60%. If both the left and right switches are pressed twice, the probability that "U" is pressed may be 0% as in the previous example, or may not be.

The sensor key portion according to the second aspect has a shape as shown in FIGS. These shapes are intended to facilitate the pressing of the sensor key portion according to the first aspect.

The shape of the sensor key portion illustrated in FIG. 5 is an example whose purpose is to press the sensor key portion to which three characters are assigned with a finger. When a specific position of the sensor key portion is pressed by a finger, it is desirable that the position has a convex shape. FIG. 5 shows a case where three character strings are arranged in a line, and FIGS. 6 and 7 show a case where three characters are arranged in a triangular shape. FIG. 8 shows a case where four characters are arranged in a quadrangular shape. The shape of the sensor key portion may be any shape such as a circle, an ellipse, a triangle, a rectangle, and a square. Further, the number of characters assigned to the sensor key portion is also arbitrary.

The shapes of the sensor keys shown in FIGS. 9 to 12 are intended for input by a stylus. 9 to 12 correspond to FIGS. 5 to 8, respectively. The shape of the sensor key portion is concave near each character so that the tip of the stylus can be pressed just above each character assigned to the sensor key portion. In this manner, by depressing the vicinity of each character, when the stylus presses the vicinity of the character to be input, the tip of the stylus slides toward the center of the character. Therefore, if the stylus presses the vicinity of the character to be input, the character to be input can be accurately pressed. Here, too, the shape of the sensor key portion is adapted to the arrangement of characters assigned to the sensor key portion.

The shapes of the sensor keys shown in FIGS. 13 to 16 are intended for input by both a finger and a stylus. 13 to 16 correspond to FIG.
8 to FIG. Here, a depression matching the shape of a finger as shown in FIG. 5 for input by a finger and a stylus as shown in FIG. 9 sliding toward a character to be input as shown in FIG. 9 for input by a stylus. It also has a shape.

In the above description, a symmetrical shape is given as an example. However, an asymmetrical shape may be adopted depending on the arrangement of the sensor key portion.

The keyboard unit according to claim 4 (FIG. 17).
Is intended to make it easier to press the sensor key section on a mobile phone or a mobile terminal. In a mobile phone or the like, the sensor key portion is often pressed with the right thumb, but if the sensor key portions are arranged radially from the right thumb, a desired position on the sensor key portion is easily pressed. FIG. 17 illustrates a sensor key portion in which characters are linearly assigned as an example, but the same idea can be applied to a sensor key portion of another shape in which characters are not linearly arranged. For example, if the sensor key portion to which three characters are assigned has a triangular shape, the triangle of each sensor key portion may be arranged so that the base of the triangle is at a right angle from the position of the thumb,
When the sensor key portion has an L-shape, either side of the L-shape may be arranged at a right angle from the position of the thumb. In addition, if the sensor key portion is arranged radially from the vicinity of the thumb, the width between the sensor key portions near the thumb becomes narrow, so the shape of the sensor key portion is made thinner near the thumb and wider when far from the thumb. do it. In FIG. 17, it is assumed that the sensor key is pressed with the thumb. However, even in the case where the sensor key is pressed with a finger other than the thumb, the sensor key may be radially arranged as in FIG. It is possible. Further, in a keyboard section that is assumed to press the sensor key section with a plurality of fingers, it is desirable to arrange the sensor key sections in a radial pattern with the positions of the plurality of fingers as the centers (FIG. 18).

As a keyboard portion in which other sensor key portions are arranged, there is a keyboard portion in which each sensor key portion is attached diagonally (FIGS. 19 and 20). In FIG. 19, when pressing both ends of each sensor key unit, the finger may be moved right and left. In FIG. 20, when pressing both ends (at a position higher than the center portion) of each sensor key portion, it is difficult to press the center (at a position lower than both ends) of the adjacent sensor key portion. There is less pressing.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a character string input device of the present invention.

FIG. 2 is a configuration diagram of a character string input device according to the present invention.

FIG. 3 is a sensor key unit.

FIG. 4 is a sensor key unit

FIG. 5 is a laterally long sensor key portion having a shape suitable for pressing a specific place with a finger.

FIG. 6 shows a triangular sensor key portion having a shape suitable for pressing a specific place with a finger.

FIG. 7 is a circular sensor key portion having a shape suitable for pressing a specific place with a finger.

FIG. 8 is a square sensor key portion having a shape suitable for pressing a specific place with a finger.

FIG. 9 is a laterally long sensor key portion having a shape suitable for pressing a specific place with a stylus.

FIG. 10 shows a triangular sensor key having a shape suitable for pressing a specific place with a stylus.

FIG. 11 shows a circular sensor key having a shape suitable for pressing a specific place with a stylus.

FIG. 12 is a square sensor key portion having a shape suitable for pressing a specific place with a stylus.

FIG. 13 is a laterally long sensor key portion having a shape suitable for pressing a specific place with a finger or a stylus.

FIG. 14 shows a triangular sensor key portion having a shape suitable for pressing a specific place with a finger or a stylus.

FIG. 15 shows a circular sensor key having a shape suitable for pressing a specific place with a finger or a stylus.

FIG. 16 is a square sensor key portion having a shape suitable for pressing a specific place with a finger or a stylus.

FIG. 17 shows a keyboard unit in which the arrangement and orientation of the sensor key unit are made easy to press with one finger.

FIG. 18 shows a keyboard unit in which the arrangement and orientation of the sensor key unit are made easy to press with two fingers.

FIG. 19 is a keyboard section in which the direction of a sensor key section is oblique.

FIG. 20 is a keyboard section in which the direction of a sensor key section is oblique.

FIG. 21 shows an example in which five pseudonyms are assigned to one sensor key unit.

FIG. 22 shows a conventional character string input device.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 Switch part 2 Sensor key part 3 Keyboard part 4 Probability estimation part 5 Language model part 6 Character string estimation part 7 Key part 8 Left switch part 9 Right switch part 10 Switch part (pressure which can detect the pressure of several places) 11 convex part 12 concave part 13 thumb 14 finger

Claims (3)

[Claims] An apparatus has one or a plurality of pressure sensors, angle sensors, switches, etc. to which a plurality of characters are assigned (hereinafter, collectively referred to as a switch unit; FIG. 1-1), and indicates which part of a key is pressed. , Or a sensor key unit (FIG. 1-2) capable of detecting the inclination of the key, and a keyboard unit (FIG. 1-3) including a plurality of sensor key units; A probability estimating unit (FIG. 1-4) that obtains and stores the probability that each character has been input for each character assigned to the sensor key from the pressed location on the sensor key unit, A language model unit (FIG. 1-5) for storing a language model, etc .; information on which sensor key unit has been pressed from the probability estimating unit; and a character with input probability information of each character assigned to the sensor key unit. Column A character string estimating unit (FIG. 1-6) for estimating an input character string from information and linguistic knowledge such as surrounding context, word constraints, language constraints, empirical language knowledge, and statistical language models of the language model unit. String input device with a). 2. A character string input device according to claim 1, further comprising: a sensor key portion having a shape which makes it easy to estimate which place is pressed or at what angle. 3. The character string input device according to claim 1, further comprising a keyboard having a sensor key portion suitable for inputting the character string.