EP0157838A1

EP0157838A1 - Reproduction of complex character symbols by combining sub-symbols.

Info

Publication number: EP0157838A1
Application number: EP84903620A
Authority: EP
Inventors: Zwi Barnea
Original assignee: Individual
Current assignee: Individual
Priority date: 1983-10-06
Filing date: 1984-10-05
Publication date: 1985-10-16
Also published as: WO1985001475A1; JPS61501382A; KR850700022A; DE3478099D1; EP0157838B1; EP0157838A4

Abstract

Une méthode de spécification des caractères de langues telles que le chinois, le japonais et le coréen, n'utilisant pas plus de quatre touches d'un clavier, chacune d'elles représentant un élément, chaque élément représentant au moins une composante d'un caractère et, en combinaison, étant capable de représenter la totalité de n'importe quel caractère de l'ensemble de caractères, lorsqu'il est introduit dans l'ordre dans lequel les composantes du caractère seraient écrites de manière conventionnelle. La méthode fournit également des moyens permettant d'identifier et de réduire au minimum la disruption d'introduction provoquée par des ensembles de caractères qui sont ambigus lorsqu'ils sont introduits de la manière décrite ci-dessus.A method of specifying characters for languages such as Chinese, Japanese, and Korean, using no more than four keys on a keyboard, each of which represents an element, each element representing at least one component of a character and, in combination, being capable of representing the totality of any character in the character set, when introduced in the order in which the components of the character would be written conventionally. The method also provides means for identifying and minimizing introductory disruption caused by sets of characters which are ambiguous when introduced as described above.

Description

Reproduction of Complex Character Symbol s by Combining Sub-Symbol s .

This invention relates to improved character manipulation and, in particular to improved means of specifying and forming characters.

The general area to which the invention relates is set out in British paten application No. 80.39683, entitled "Method and Apparatus for Specifying an Forming Characters" of the present applicant and Shui-yin Lo (corresponding U.S. patent application serial No. 215552, now replaced by Continuation-in- Part application No. 476347).

This invention described the use of a limited number of elements to represent each part of a Chinese character, the elements each being given an alpha¬ numeric code and a character being defined by the codes of the elements forming the character in the order in which they would be written.

That invention then describes the method of identifying the particular characters from a dictionary, displaying these and, if required, providing a hard copy output.

That specification also describes certain methods whereby ambiguities can be resolved.

In a later application by the same applicants, British patent application No. 83.07989, entitled "Character Manipulation", the subject matter of which was also incorporated in the United States Continuation-in-Part application referred to above, there were disclosed an ergonomic form of keyboard and also new methods whereby certain ambiguities can be removed.

U.S. patent No. 4,379,288 of Leung discusses, historically, developments in specifying Chinese characters.

The specifications of the two prior applications of Barnea and Lo can be read in conjunction with this specification and, generally, I shall not describe herein material which has already been described in those two prior specifications.

The object of the present invention is to provide various improvements in the concepts set out in the earlier applications.

In the first aspect of the invention, I have discovered that the number of elements necessary for specifying characters can be less than hitherto thought and I have found, surprisingly, that the reduction of the number of elements to four or even three introduces few additional ambiguities, whilst effectively saving half the useable memory space required to specify the character set in the computer with which the invention is used.

In this aspect I provide a method of specifying characters by the use of not more than four keys on a keyboard, each of which keys represents an element, each element representing at least one component of a character and which, in turn, is represented by an alpha-numeric code, the elements, in combination, being capable of representing the whole of any character of the character set, when entered in the order in which the components of the character would conventionally be written.

The actual keyboard could, if required, have only the four keys necessary for the application of the method but, practically, the keyboard may have a number of other keys which could be used for frequent element combinations which are ergonomically located on the keyboard.

These element combinations may comprise radicals or characters but do not necessarily comprise these.

The location of the various combinations can be defined as described in British patent application No. 83.07989 previously referred to.

In a second aspect of the invention, I have found that it is possible to less ambiguously define complex characters by omitting to individually specify certain elements, often intermediate elements, by the use of an "empty" key, whereby an unspecified number of elements is omitted, and then complete the specification by the use of the final elements. Alternatively, the "empty" key can provide truncation or omission to save input effort.

In a further aspect of the invention, I have found that where specific information relating to the relative position of strokes is required to avoid ambiguity, particularly amongst simpler characters, by providing an alternative method of input, possible ambiguities are reduced.

A still further aspect of the invention is the division of the ambiguities into two sets: a set of the most common ambiguities and a set of less common ambiguities. The most common set would be memorised by the operator. The first ambiguity of each ambiguity group (consisting most commonly of two or three characters) would be specified by its usual input and need not be memorised. The other member or members of the group would be specified by the operator by an additional keystroke or by a specified sequence of keystrokes.

A still further aspect of the invention is in the use of the keyboard to abbreviate the most common characters by representing them by single keystrokes which, on their own, do not represent characters, or by combinations of two keystrokes which cause the input for these characters to be abbreviated and are not otherwise utilized. These can have high mnemonic values. Combinations of two keystrokes can also be used to represent long initial radicals of characters and thus facilitate the entering and differentiation of such characters.

In order that the invention may be more readily understood, I shall more fully describe each of these aspects.

The following description will be in respect of Chinese characters, but it is to be appreciated that the invention has application in other languages using ideographs, specifically Japanese and Korean, which can be considered to be made up of a small number of elements.

In the earlier of the two applications of the present applicant and S-y Lo, the preferred embodiment indicated that characters could be described by the use of five elements, each of which was given an alpha-numeric code, the elements being a short stroke or dot, a right sloping diagonal, a left sloping diagonal, a horizontal and a vertical line.

I have found, most surprisingly, that if, instead of using five elements, by using not more than a foursome, I can provide a method which is effectively as satisfactory as that described in the earlier application, as far as differentiating characters is concerned, but which provides a remarkable saving on computer memory space.

It will be appreciated that, in order to provide a binary code for five different elements, it is necessary to use three bits of information and, as most computers operate in multiples of four bits, it becomes practically necessary to allot four bits for each element. Where a computer memory is to hold particulars relating to a large number of characters, and some of these have a substantial number of elements, this can lead to the 'waste' of a substantial part of the computer memory.

On the other hand, where four or less elements are being used, then only two bits of information are needed for each element and this enables four elements to be defined by each eight bit word. Thus, there is, effectively, a doubling of the utilisation of the memory needed for coding of characters. This makes a very substantial difference where a library of several thousand characters is to be maintained in ROM and is particularly substantial if account is taken of the totality of data stored and communicated in a large country.

The advantage obtained, whilst very real, would, of course be negated if the reduction in the number of elements led to a great increase in ambiguities introduced.

When I decided that it would be desirable to restrict the number of elements to not more than four, I first considered, logically, those elements which did not normally appear in similar groups of elements in any substantial number of characters.

Initially I considered foursomes, and intuitively, I came to the conclusion that the short stroke and the right sloping diagonal basically satisfy this criterion and, as such, I allotted each of these the same code and, from reference to the attached Figure 1, it will be seen that the arrangement adopted, set 1 in the Figure, provides short strokes, stops, right diagonals are all allotted code 1 and the other three members, the horizontal line, the vertical line and the left diagonal are 2, 3 and 0 respectively.

For such an arrangement to be satisfactory, it must be seen whether the particular foursome met criteria which I believe are essential.

These are:

1. That all the features of any character can be represented in terms of one or other of members of the foursome.

It appears clear that the particular foursome selected does, in fact, meet this criterion.

2. That the number of ambiguities be not excessive.

This will be discussed further later herein, but suffice to say that the adoption of this particular foursome shows no impractical increase in the number of ambiguities over those found in a five element set.

3. That the set be intuitively acceptable.

This goes basically to operator effectiveness.

If an operator cannot understand why particular components are selected or finds difficulty in actually ascertaining which component should be represented, then the use of the foursome will be inefficient and unsatisfactory. If, on the other hand, the elements selected are logical and obvious, then an operator will have no difficulty, intuitively, in applying the rules necessary to select the form of element.

4. That the elements set must be economical with respect to computer storage.

This, of course, will be satisfied by any foursome or lesser number of elements selected.

Whilst using the foursome of set 1 of the attached Figure, I have ascertained that, in the most frequently used 3700 characters in Chinese, the reduction of the number of elements from the previous five to the foursome of set 1 of the Figure introduces only one ambiguity pair which did not occur using five elements. It is confidently expected that this number of ambiguities will not increase substantiaEy with any expansion of the character set.

Thus, the advantages gained by halving the required memory space, more than offset the disadvantage introduced by this reduction in the number of elements.

After carrying out tests with the foursome of set 1 of the Figure, I then considered whether or not any other foursome would be equally or substantially equally as satisfactory.

The foursomes of sets 2, 3 and 4 of the Figure, show different alternatives. By way of explanation, the letters CCW and CW in the foursome of set 2 indicate curved elements which are counter-clockwise and curved elements which are clockwise.

It will be appreciated that the clockwise elements not only include the left diagonal stroke, but also any strokes which generally curve in a clockwise direction. Similarly, the anti-clockwise elements will include not only the right diagonal stroke, but also any elements which move in a generally anti¬ clockwise direction. In this case, by definition, the short stroke or dot must be incorporated in one of the elements concerned and I would normally prefer to count this as a counter-clockwise element but it is stressed that this is purely for convenience.

OMPI

1/4 WO A* Also, if required, by adopting a convention, any very commonly occurring element or radical could be incorporated into one of the elements. For example, I have indicated in the foursome of set 2 that the vertical stroke and the mouth radical could be arbitrarily given the same coding.

The foursomes of sets 3 and 4 are simply exemplary of particular further combinations of elements which could be used to make up a foursome which basically satisfies the criteria set hereinabove.

Somewhat surprisingly, further consideration led to the threesome of set 5, which appears to substantially satisfy the criteria set out hereinbefore.

Another advantage which may be gained by selecting particular foursomes or threesomes where the individual elements of the sets can, themselves, be non- unique can be that particular radical keys which may otherwise be provided on the keyboard but which are incorporated as an element of the set can be omitted and, thus, the key is freed for a further purpose, such as a more complex radical.

Also, if, instead of attempting to achieve maximum utilisation of the gained memory, one would rather attempt to decrease ambiguities generally, it will be appreciated that it is possible to store less truncated versions of the character codes and, thus, decrease the number of ambiguities due to truncation.

I have found that, for each additional digit encoded, the number of ambiguities decreases by one half. Thus, by the use of a foursome, a compromise can be reached which provides improved memory usage whilst, at the same time, provides a positive result in that the total number of ambiguities in normal operation is reduced.

Whilst I have indicated that only four or three rather than five elements are used to specify the characters, it would, of course, be possible to still use a five element keyboard but with the numerals representing the short stroke and right sloping diagonal being the same.

Thus, the new method could be completely transparent to an operator-used to operating under the five element system, but the benefits, as far as memory usage would still be achieved.

In the second aspect of the invention, I have found that many long characters are overspecified if their total element sequence is used and can be positively recognised even where some of the elements, particularly some of the intermediate elements, are omitted.

Purely for example, all metals have the character for gold as a prefix and particular specifying components to indicate the particular metal and many characters to do with art have the character for silk as a prefix, and have special suffix components to define the particular aspects.

I have found that when introducing the first elements and then leaving a gap, of an indeterminate number of elements in continuous sequence, and then completing the elements of the character, ambiguities are few and may be substantially less than in the situation where an input which comprises a specified arbitrary number of elements, in sequence, is used.

In order to obtain this result, I provide a specific key, an "empty" key, which provides an indication that there is a continuous sequence of unspecified elements omitted so that, when searching its memory, the computer looks for a string equivalent to the elements initially entered and then checks, in the characters having this initial string, if the elements entered after the "empty" key are present.

It is preferred that a complete radical not be omitted and it may be practicable to permit the use of the empty key in more than one position in a particular character which would, of course, necessitate effectively a double string search after the initial entry and an ascertainment that the members of each string do not overlap.

This feature is basically useful where more complex characters are being examined and, should it leave ambiguities, then these will be displayed in the manner as described in the earlier specification.

The "empty" key can be used in three different areas.

_ O 3MMPPII iVIPO The first of these is effectively that referred to above, that is to avoid a substantial part of a long initial radical which is common to a large number of similar characters.

The second is as a truncating device to simply abbreviate the input by finishing inputting after a particular number of elements but, nevertheless, indicating that the character itself does not have only that number of elements.

A third area where the use of an "empty" key can be of value is where there is an unclear, if copying from handwriting, or uncertainly remembered, portion of a character. In this case, the recognised or known portions of the character can be entered and the unclear or uncertainly remembered portion can be omitted by the "empty" key and, either, the whole character will be displayed or the normal method of resolving ambiguities can be effected if more than one character includes the elements entered and other elements in the places where the "empty" key has been used.

The economy achieved by the judicious use of the empty key can be substantial, especially when used with the traditional and more complicated characters used in Hong Kong, Taiwan, Korea and Japan.

In the third aspect of the invention, I provide alternative means to overcome the ambiguities involved in the use of relatively short characters.

As will be appreciated from the earlier specifications, in linearising a two dimensional character into a string of alpha-numeric codes, important differentiating information about the relative positions of the strokes is disregarded and lost. This, of course, was accepted as a disadvantage of the earlier system and means were provided to overcome the ambiguities caused by this.

I have found that it is relatively straightforward, as far as short characters are concerned, to require the keys to be used in a structured way to avoid these ambiguities at the time of input, rather than to distinguish between them after the ambiguity has been inputted. Purely for example, the character for man Λ would normally be entered by the typing of two separate keys J and v» (0 and 1) but, because it is a relatively regularly used combination, there is a single key provided for the combination . On the other hand, the character A also requires an input of the keys 0 and 1. If then, by convention, the man character was always required to be typed using the single key and the other character by using the two separate element keys equivalent to 0 and 1, then the ambiguity between the two would be automatically removed.

In order to assist an unskilled operator, it might be possible to still show an ambiguity when the separate keys 0 and 1 are entered, but, in this case, to show the least common usage (or usages) as the preferred usage, on the basis that the other usage would normally be effected by the use of a single key.

In a second example, the character _ would be entered as 20 U and the character * ± entered as 203 ±. as the components O and ___. are sufficiently common to be marked on the keyboard.

In an alternative, where particular components are shown on the keyboard, in the case of short characters having, for example, at most eight elements amongst which characters most ambiguities occur, the arrangement could be that some or all of the keys have the literal meaning marked on the key, when the key is operated, but, if there is another character component which uses the same digit combination, the use of a shifted key could indicate the alternative form or forms which are less common than the one shown on the key concerned, or could indicate the size or relative position of a component.

For example, the two characters o and jfc. would both be entered, if the element keys were used, as 3232 but, using this aspect of the invention, the first form would be obtained by using simply the key bearing the marking and the second by pressing the same key as a shifted key. Of course, if there were a number of ambiguities, the use of the shifted key would still show the alternatives, either with the most common omitted, or preferably, the less likely in the circumstances, as normally the most likely would have been obtained by using the unshif ted key.

It will be seen that, by the use of this aspect of the invention, a skilled operator can limit the number of ambiguities which are found on the screen, whilst an unskilled operator can still obtain a result from basics although, possibly, the separation of ambiguities would be slightly more tedious than in the method described in the earlier of the prior applications, in that the most likely character would normally be displayed as the least likely in the listed ambiguities.

A still further alternative method of resolving ambiguities is based on the knowledge that in the first thousand most commonly used characters, I have found that there are, in fact, only fifty-five ambiguities and it will be appreciated, from studies of the frequencies of characters, that the first thousand most common characters would include 95% of the characters of an average text. It is possible to memorize these most common ambiguities. Ambiguities of lesser frequency than the most common set, of say fifty-five ambiguities, would be signalled and resolved by the operator.

It is possible to establish an arrangement whereby, where the existence of an ambiguity set is known, simply entering the elements of the character can display the character which is the most common application of the element set but, entering the elements and then, say, pressing the last key twice, the character which is the second most common application of the element set will be displayed and effecting the pressing of the specific key three times would show the character which is third most common character, and so on, a skilled typist, while learning possibly only thirty-five specific ambiguity situations, can automatically initially adopt the correct character for a particular element string and, thus, not have to go through the procedure of resolving ambiguities in these learned common characters. Alternatively, the specification of which character of an ambiguous set is intended can be effected by some other key on the keyboard, such as, for example, the ambiguity resolution key.

Memorization of the most common ambiguities can be simplified by various mnemonic devices including differentiation by describing the different sound values of ambiguous characters.

This serves a substantial purpose in that the operator's attention is not continually diverted to resolve ambiguities which, in most cases, will mean

O PI adopting the particular character initially found and, by using such an arrangement, I find that I can limit the ambiguity rate in an average text to approximately 0.3% or one in three hundred characters.

It would, of course, be possible to provide an alternative mode for an unskilled operator to simply enter the elements of the character in the normal way and then to enter the normal ambiguity resolution process.

In a still further embodiment of the invention, which would be most useful for persons who are continually using the method of the invention, I may provide a means whereby a specified input, which may or may not be relative to the actual character required, consisting of one or only a small number of key strokes can be used, by definition, to display certain characters. This could be done by associating various characters with any radical key, or other key on the keyboard which would not normally be used alone, or any two such keys to indicate that a specific learned character is required. It may be possible to select the key or keys used to be a mnemonic of the required character.

It would also be possible to use such a key or keys to define a common, complex initial radical to minimise the entry requirements so that the differentiating elements can be entered after few keystrokes.

Claims

CLAIMS:

1. A method of specifying characters by the use of not more than four keys on a keyboard, each of which keys represents an element, each element representing at least one component of a character and which, in turn, is represented by an alpha-numeric code, the elements, in combination, being capable of representing the whole of any character of a character set, when entered in the order in which the components of the character would conventionally be written.

2. A method as claimed in claim 1 wherein there are four keys and the elements constitute a foursome.

3. A method as claimed in claim 1 wherein there are three keys and the elements constitute a threesome.

4. A method as claimed in claim 1 wherein each of the elements is unique.

5. A method as claimed in claim 1 wherein each alpha-numeric code can represent more than one component.

6. A method as claimed in claim 2 wherein the foursome consists of a diagonal line, in one sense, and short strokes; a horizontal line; a vertical line; and a diagonal line, in the second sense.

7. A method as claimed in claim 2 wherein the foursome consists of all components which are formed counter-clockwise; a horizontal line; a vertical line; and all components which are formed in a clockwise manner.

8. A method as claimed in claim 7 wherein short strokes are included with the counter-clockwise components.

9. A method as claimed in claim 7 or claim 8 wherein the mouth radical is associated with the vertical line.

10. A method as claimed in claim 1 wherein, on entering a character, after entering certain elements, the existence of a certain specified group of elements is indicated by the use of an "empty" key, the characters having the initial elements being searched.

11. A method as claimed in claim 10 wherein the "empty" key is used at the end of the inputted characters to indicate truncation.

12. A method as claimed in claim 10 wherein when certain elements are entered after the use of "empty" key, these elements are the subject of a string search.

13. A method as claimed in claim 10 wherein the "empty" key is used more than once to indicate the existence of more than one group of unidentified elements, the characters having the initial coding being searched to ascertain if they have the groups of elements, then the group of elements subsequent to the first of the "empty" key is searched for matching.

14. A method as claimed in claim 12 wherein, where there are at least two other groups of elements, the search ensures that both of the groups are present and that these are exclusive relative to each other.

15. A method as claimed in claim 1 wherein the keyboard has, as well as the element keys, a number of component keys each of which represents a number of elements, in a particular order, which keys are ergonomically located on the keyboard about the element keys.

16. A method as claimed in claim 15 wherein where a particular combination of elements can be achieved from the use of the element keys or at least one component key, with element keys if required, the specific keys used to specify a character can be used to differentiate between a known ambiguity or ambiguities.

17. A method as claimed in claim 15 wherein, to reduce ambiguities, component keys or the individual keys to produce the same element combination as the component can be selectively used to select between a known ambiguity or ambiguities.

__ O PI

18. A method as claimed in claim 17 wherein, when the element keys are used, an ambiguity is shown but the second most common member of the ambiguity set is displayed.

19. A method as claimed in claim 15 wherein the entry of a component key represents the most common form of the component and the use of the shifted or otherwise selected component key represents the next most common use of the component.

20. A method as claimed in claim 15 wherein, after entering the elements of a character known to be ambiguous, a specific key can be appropriately pressed to indicate an ambiguity is known and the required member of the ambiguous set is selected.

21. A method as claimed in claim 19 wherein the specific key can be pressed more than once, depending upon the required ambiguity in a set, and, on entry of the input, the required character will be displayed.

22. A method as claimed in claim 15 wherein where the component keys do not represent a character, the component keys can be used alone or in arbitrary or in mnemonic combination, to represent specific alpha¬ numeric codes, and thus characters, without the characters having to be defined by the use of the element keys or other component keys on the keyboard.

23. A method as claimed in claim 22 wherein the component keys used alone or in combination can represent a complex radical.