T 0922/04 (Disambiguating system/TEGIC COMMUNICATIONS) of 9.11.2007

Summary of Facts and Submissions

I. European patent No. 0 842 463 concerns a reduced keyboard disambiguating system, the patent claiming priorities from patent applications filed in 1995 and 1996. Claim 1 of the European patent reads as follows:

"An input disambiguating system (50) for disambiguating ambiguous input sequences entered by a user, the disambiguating system comprising:

(a) a user input device (54) having a plurality of inputs (56), each of the plurality of inputs being associated with a plurality of characters, an input sequence being generated each time an input is selected by manipulating the user input device and corresponding to the sequence of inputs that have been selected, the generated input sequence having a textual interpretation that is ambiguous due to the plurality of characters associated with each input;

(b) a memory (104) containing a plurality of objects, including word objects comprising textual objects that are completed words and word stem objects comprising textual objects that comprise a string of characters corresponding to the initial characters of an uncompleted word, each of the plurality of objects being associated with an input sequence;

(c) a display (53) to depict system output to the user; and

(d) a processor (100) coupled to the user input device, memory, and display, the processor identifying from the plurality of objects in the memory at least one object associated with each generated input sequence, and generating an output signal causing the display to display at least one of the identified objects associated with each generated input sequence as a textual interpretation of the generated input sequence."

II. The patent was opposed and finally revoked on grounds of lack of novelty and inventive step by the opposition division in a decision posted on 11 May 2004. The revocation was based on prior art disclosing input devices for alphabetic writing systems. The opponent also cited prior art concerning logographic writing systems, in particular the document

D16: US-A-5 109 352 (published on 28 April 1992)

concerning a system for encoding the Chinese and Japanese scripts.

This group of prior art documents was rejected as not relevant to the invention. According to the decision under appeal, Roman letters as used in the English language were not equivalent to the strokes making up a Chinese character, an argument suggested by the opponent. Whereas in the English language the sequence of letters defined a word, it was the two dimensional spatial relationship of strokes relative to each other which determined a Chinese character. The order in which the strokes of a Chinese character had been drawn was not visible and was thus not a characterising feature like the order of Roman letters displayed in an English word. Similarly, the decision also denied the equivalence between Chinese radicals and the word stem objects defined in claim 1 as suggested by the opponent.

III. The patent proprietor filed an appeal against the revocation decision on 9 July 2004, paying the appeal fee on the same day, and filed a written statement of the grounds of appeal on 14 September 2004. Together with the statement of grounds as well as by a further letter dated 2 December 2005, the appellant filed requests for amended claims. The respective first independent claim of these requests was amended in respect of claim 1 as granted as follows (underlining added to indicate amendments; unamended passages are omitted as indicated by ellipses):

Main request and 2nd, 3rd, and 4th auxiliary requests:

"An input disambiguating system […]

(a) […]

(b) a memory (104) containing a plurality of objects including completed word objects [… ;] and each object belonging to one of a plurality of object types;

(c) […]

(d) [… .], the system being arranged such that the display of word stem objects during entry provides feedback to the user confirming that the correct sequence of inputs has been entered to lead to the entry of the disambiguated input sequence intended by the user."

1st and 5th auxiliary requests:

"An input disambiguating system […]

(a) […]

(b) […]

(c) […]

(d) [… .], and wherein each of the plurality of objects in the memory is associated with a frequency of use, wherein the system changes the frequency of use to reflect a frequency with which a user uses certain objects within the memory."

6th auxiliary request:

"An input disambiguating system […]

(a) […]

(b) a memory (104) containing a plurality of objects including completed word objects [… ;] and each object belonging to one of a plurality of object types;

[…]

(d) [… .], wherein the identified objects associated with the input sequence are displayed in a selection list on the display, and wherein if a plurality of objects comprising two or more object types are identified the selection list comprises identified objects of a plurality of object types."

7th auxiliary request:

"An input disambiguating system […]

(a) […]

(b) […] and in which each of the plurality of objects in the memory belong to one of a plurality of object types, including a completed word type comprising textual objects that are completed words and a word stem type comprising textual objects that are the initial characters of an uncompleted word;

(c) […]

(d) [… .], wherein the identified objects associated with the input sequence are displayed in a selection list on the display, and wherein if a plurality of objects of both a word object type and a word stem object type are identified the output signal causes the display to display objects of both those types."

8th auxiliary request:

"An input disambiguating system […]

(a) […]

(b) [… ;], wherein each of the plurality of objects in memory belong to one of a plurality of object types, including a completed word type comprising textual objects that are completed words and a word stem type comprising textual objects that comprise a string of characters corresponding to the initial characters of an uncompleted word;

(c) a display (53) to depict system output to the user in response to user manipulation of said user input device; and

(d) [… .], wherein the identified objects associated with the input sequence are displayed in a selection list on the display, and wherein if a plurality of objects of both a word object type and a word stem object type are identified the output signal causes the display to display objects of both those types and further wherein each of the plurality of objects in the memory is further associated with a frequency of use, wherein the system changes a frequency of use, wherein the system changes a frequency of use to reflect the frequency with which a user uses certain objects within the memory, and wherein the identified objects associated with the same input sequence are presented to the user in order of decreasing frequency of use."

IV. The respondent (opponent) filed a written statement in response to appellant's requests, citing the following additional prior art:

D29: GB-A-2 197 097 (published on 11 May 1988)

D30: H.S. Venkatagiri: "Efficiency of Lexical Prediction as a Communication Acceleration Technique", AAC Augmentative and Alternative Communication, Volume 9, September 1993, pp. 161 to 167

D31: J.L. Arnott et al.: "Probabilistic Character Disambiguation for Reduced Keyboards Using Small Text Samples", AAC Augmentative and Alternative Communication, Volume 8, September 1992, pp. 215 to 223

V. In summons to oral proceedings, the Board indicated its intention to consider documents D29 to D31 and raised, as a provisional view, doubts regarding patentability of the claimed invention.

Oral proceedings before the Board were held on 9 November 2007. The matter in issue was discussed with the representatives of the parties on the basis of the requests presented to the Board.

VI. According to the appellant, the claimed invention was clearly novel and inventive over the cited prior art.

The storage of both types of objects, stem and word objects, and the display of stem objects during entry in order of decreasing frequency of use to give feedback to the user, was not known from the prior art.

Document D16 and the further prior art concerning the Chinese and Japanese languages were not relevant to the invention since a logographic writing system was entirely different from an alphabetic writing system used, for example, in writing English, which was the field of the present invention. Except for some correspondence between Chinese characters and English words, there was no other relevant correspondence or equivalence between the basic writing components of the Chinese and English languages. Writing components like the strokes used in writing the Chinese characters and used in document D16 as a crucial element of the disambiguating algorithm had no function in the present invention. The strokes for composing a Chinese character did not correspond to any textual object used by the present invention; they were actually meaningless graphical units, comparable only, if at all, with the component lines of a written Roman letter.

Document D29 disclosed an algorithm for ordering a list of words according to the recency of use. Maintaining absolute word frequency values, however, was indicated as being computationally not economic.

Document D30, although addressing a system for ordering complete words according to their frequency of use and changing the frequency as words were entered, did not suggest to present word stems selected from stored word stem objects in a memory. The document did not suggest that there was any problem with presenting complete words to the user.

Document D31 only disclosed the idea of associating a frequency of use with objects in a memory and changing that frequency. As with documents D29 and D30, however, the skilled person would not combine the teaching of this document with that of the remaining prior art to arrive at a system having all of the features of the claimed disambiguating system.

VII. Accordingly, the appellant requested that the decision under appeal be set aside, and that a patent be granted on the basis of the main request or alternatively on the basis of the first, second, third, fourth, fifth, sixth, seventh and eighth auxiliary requests filed with letters dated 14 September 2004 and 2 December 2005.

VIII. The respondent requested that the appeal be dismissed.

IX. The respondent raised various objections, including added subject matter and lack of clarity, and argued that the claims under consideration did not meet the requirements of novelty and inventive step, respectively.

According to the respondent, the opposition division was wrong to disregard the input systems for Chinese or Japanese scripts. The claim wording, in particular if read in the context of paragraph [0131] of the present patent specification, did not limit the invention to the English or any alphabetic writing system at all. The opposition division had given undue emphasis to the writing system as perceived by a human user. The disambiguating system did not have any understanding of the text it writes; it rather accepted a limited number of defined inputs, performed some searching of its memory and displayed an object which corresponded to the input sequence. Such a system was not aware of language and was immune to any preconceptions about differences between written Chinese and written English. It was only concerned with the order of entry, and not at all with the final appearance of the character or word. The only precondition which both the present invention as well as the prior art required was that the order in which the basic components of a word, i.e. the strokes or letters, were drawn was fixed. This was the case in the English as well as in the Chinese writing system.

Document D16, in particular, taught that successively more complete portions of a character might be displayed each time a stroke was entered, causing the display to render at least one of the identified objects associated with each generated input sequence as required by the present invention.

The new requests introduced the additional feature of associating with the objects, a frequency of use, which was previously not the subject of any claim and had thus not been searched. This meant that prior art not previously relevant, namely documents D29 to D31, now had to be taken into account. Documents D29 and D30 showed the concepts of associating a changing frequency of use with words in a dictionary. Document D31 taught applying such a frequency of use to a reduced keyboard entry system.

X. The Board announced the decision at the end of the oral proceedings.

Reasons for the Decision

1. The appeal complies with the requirements of Articles 106 to 108 and Rules 1(1) and 64 EPC and is thus admissible.

2. The appeal, however, is not allowable since the respective first independent claim of the present requests seeks protection for subject matter which lacks either novelty or inventive step and does thus not comply with the requirements of Article 52(1) EPC.

3. These claims are directed to embodiments of different aspects of the same invention and can thus be brought together under a single combined definition and considered in common without affecting the substance of the invention and the outcome of the examination. The following listing summarises the combined subject-matter of the claims (the claim features are renumbered for convenience):

(1) A disambiguating system comprising a user input device, a display, a memory, and a processor coupled to the user input device, memory and display;

(2) the user input device having

(21) a plurality of inputs, an input sequence being generated each time an input is selected by manipulating the user input device and corresponding to the sequence of inputs that have been selected,

(22) each of the plurality of inputs being associated with a plurality of characters, the generated input sequence having a textual interpretation that is ambiguous due to the plurality of characters associated with each input;

(3) the display suitable to depict system output to the user in response to user manipulation of said user input device;

(4) the memory (104) containing a plurality of objects,

(41) each of the plurality of objects being associated with an input sequence;

(42) wherein each of the plurality of objects in the memory belong to one of a plurality of object types; including

(421) a completed word type comprising textual objects that are completed words and

(422) a word stem type comprising textual objects that are (comprise a string of characters corresponding to) the initial characters of an uncompleted word;

(43) wherein each of the plurality of objects in the memory is further associated with a frequency of use,

(44) wherein the system changes a frequency of use to reflect the frequency with which a user uses certain objects within the memory,

(5) the processor identifying from the plurality of objects in the memory at least one object associated with each generated input sequence, and

(6) generating an output signal causing the display to display at least one of the identified objects associated with each generated input sequence as a textual interpretation of the generated input sequence,

(61) such that the display of word stem objects during entry provides feedback to the user confirming that the correct sequence of inputs has been entered to lead to the entry of the disambiguated input sequence intended by the user,

(62) wherein the identified objects associated with the input sequence are displayed in a selection list on the display, and

(63) wherein if a plurality of objects of both a word object type and a word stem object type are identified the output signal causes the display to display objects of both those types and further

(64) wherein the identified objects associated with the same input sequence are presented to the user in order of decreasing frequency of use.

The first independent claims of the main request and the second, third and fourth auxiliary requests are identical and comprise features (1) to (42) and (5) to (61), to which - apart from minor reformulations - the remaining auxiliary requests add the following features, respectively:

- sixth auxiliary request: features (62) and (63),

- seventh auxiliary request: features (421), (422), (62) and (63),

- first and fifth auxiliary requests: features (43) and (44), and

- eighth auxiliary request: features (421), (422), (43), (44), (62), (63) and (64).

4. The main request as well as the second, third, fourth, sixth, and seventh auxiliary requests do not meet the requirement of novelty, Article 52(1) and 54 EPC, since all features of the respective first independent claim of these requests, i.e. the features (1) to (422) and (5) to (63) above, are anticipated by document D16.

4.1 In the terminology of the present claims, document D16 discloses a disambiguating system (computer input system for Chinese and Japanese characters) comprising a user input device, a display, a memory, and a processor coupled to the user input device, memory, and display (see document D16, fig. 4: keyboard 22, monitor 30, storages 26 and 28, and controller 24).

The user input device has a plurality of inputs (keys, see D16, fig. 6), each of the plurality of inputs being associated with a plurality of characters (group of strokes, see D16, figures 1A,4, and 6 with specification, col. 8, lines 24 to 43, col. 9, lines 41 to 54, and col. 10, lines 43 to 47) which can be selected by manipulating the user input device. This produces an input sequence (string of code numbers, see D16, col. 9, lines 44 to 50; data strings, see fig. 5 and specification col. 10, lines 25 to 29) having a textual interpretation (partial character, root, and character, see col. 5, line 55 to col. 6, line 11, col. 12, lines 23 to 31) that is ambiguous due to the plurality of characters associated with each input (see D16, col. 10, lines 49 to 59, for example).

The display depicts system output to the user in response to user manipulation of the user input device (see D16, fig. 10 and specification, col. 11, lines 65 ff.).

4.2 Furthermore, document D16 anticipates features (4) to (422) (numbering, see above): The memory contains a plurality of objects (shapes of characters or partial characters, see D16, col. 9, lines 50 to 61, col. 12, lines 61 to 67, col. 21, line 17 f. and col. 24, claim 28) being associated with the input sequence (string of code numbers, D16, loc. cit.). Each of the plurality of objects belong to one of a plurality of object types (partial characters, i.e. roots or portions of roots -- see D16, col. 12, lines 24 to 31 -- as well as characters and compounds -- see D16, col. 5, line 65 to col. 6, line 11).

These object types comprise the completed words, i.e. in document D16 the Chinese (or Japanese) characters, as well as the word stems or uncompleted words, in document D16 the partial characters (roots or portions of roots) stored and displayed when the sequence of code numbers, if entered according to the traditional stroke order, uniquely identifies the string of the respective partial character.

The association of characters with completed words in terms of the present claims was unanimously accepted as appropriate by the appellant as well as by the examining division. However, it was strongly disputed (see point VI above) that a partial character formed by a sequence of strokes as disclosed in document D16 was a word stem in terms of the present claims.

This argument stands and falls on the definition of the term "word stem", which according to its ordinary meaning refers to the unchanged part of an inflected word. The present claims use this term differently by associating it with "a string of characters corresponding to the initial characters of an uncompleted word". The patent specification, paragraph [0034] is more precise on this point, indicating that word stems "are incomplete words" and "represent[ing] all possible valid sequences of letters that a user may be entering". This means that the term should not be construed according to its ordinary linguistic meaning, but in an operative sense simply as a sequence of key codes entered by the user in the intention to write a word.

As rightly pointed out by the respondent (see point IX above), the invention is not limited to the application to an alphabetic or any other specific writing system. Therefore, the term "character" in the present claims could and should be understood in a broad sense, including Roman letters but also more basic writing components like the strokes used for composing a Chinese character.

There is no justification for a more narrow construction of the claims. In fact, a straightforward implementation of the claimed invention is a digital data processing system which encodes any input into binary zeros and ones and retrieves the shapes of the basic writing components automatically from a database, e.g. from font files or otherwise. Beyond zeros and ones, therefore, the system has no idea of language or the difference between alphabetic and logographic writing systems. It does not make any relevant difference whether the disambiguating system operates on digital data encoding Roman letters or on digital data encoding strokes of the Chinese writing system for composing the words the user intends to input. For disambiguating an ambiguous input sequence, the linguistic meaning of the input or the graphical appearance of the data to be displayed is also irrelevant provided that the system reproduces the input sequence fundamental to possible meanings and accesses the right databases. The same holds for displaying and printing the data: the system retrieves the graphical information e.g. from an appropriate font file. It does not matter for the functioning of the system whether the retrieved symbol is a Roman letter or a Chinese stroke or character or which semantic meaning the displayed symbol has. Such aspects of the language and writing system used are not essential to the present invention.

The Board concludes from these considerations that the strokes in document D16 are included in the present definition of character, and accordingly the partial characters, roots and portion of roots referred to in document D16 are also included in the present definitions of word stem and uncompleted word.

4.3 Moreover, document D16 also anticipates features (5) to (63) (numbering, see above):

The processor (controller 24) identifies at least one object in the memory and displays it as a textual interpretation of the generated input sequence (see D16, fig. 10, partial characters 58 to 68, and 56 and pending character 52 and col. 12, lines 8 to 60). The displayed object can be a word stem (partial character, i.e. a root or a portion of root), which provides feedback to the user confirming that the correct sequence of inputs has been entered (see D16, for example col. 12, line 17 f., col. 13, lines 19 to 27).

A plurality of identified objects associated with the same input sequence (number of duplications) are displayed in a selection list: "one or more and less commonly used alternative characters are shown in parenthesis alongside the most commonly used character so that the operator can make a choice by a single keystroke" (document D16, col. 16, lines 9 to 20). The display shows completed words (characters) in pending area 52, which can be moved to text line 54 by actuating the space bar, and simultaneously a word stem (56) to which further strokes may be added (D16, col. 16, line 60 to col. 17, line 16).

It follows that document D16 fully anticipates the subject matter of the first independent claims of the main request as well as of the second, third, fourth, sixth, and seventh auxiliary requests.

5. The first, fifth, and eighth auxiliary requests do not meet the requirement of the inventive step, Article 52(1) and 56 EPC, since the subject matter of the respective first independent claim is rendered obvious by document D16 and, for example, document D31.

For these claims, document D16 is certainly an appropriate starting point for assessing inventive step. The only new features over this prior art, features (43), (44), and (64) (numbering, see above), concern the presentation of identified objects according to their frequency of use.

However, already this prior art teaches to present the "most commonly used" of the identified characters to the user, which alludes to the above "frequency of use" features of the present claims at issue (see document D16, col. 16, lines 14 to 20). Documents D29, D30, and D31 share the same idea.

In particular, this last cited document brings out the "adaptive modelling of character sequences" as a means for increasing the efficiency of a disambiguation system for a reduced keyboard (see document D31, p. 222, second paragraph). According to this document, the occurrence of any word or portion of a word (n-gram) is to be counted, providing numbers which indicate the frequency of use of the words or word portions stored in the dictionary. The numbers are used as a measure of the probability of occurrence of the word or a portion of a word in predicting and presenting character sequences (see document D31, p. 217, right-hand column, last paragraph and p. 218, right-hand column, first paragraph).

Applying such an adaptive method to the disambiguating system of document D16, which directly leads to the implementation of features (43), (44), and (64), is straightforward and, in the light of document D31, an obvious measure to increase efficiency of the system. The claims at issue do thus not meet the requirement of inventive step.

ORDER

For these reasons it is decided that:

The appeal is dismissed.