Electronic dictionary capable of identifying idioms

[0002] This invention relates to an electronic dictionary, and in particular to an electronic dictionary capable of handling idioms.

[0003] For many years, people have been studying the machine translation technology which can translate written or speaking expressions from one language into another by means of computers. Machine translation is a kind of automatic translation from one language into another (or many others) without human aid. On the basis of linguistic analysis regarding the form and structure of a language, it builds machine lexicon and machine grammar with the mathematical approach, by taking advantage of the huge power of computers in storing and data processing. In order to realize the language translation, a machine translation system has to have the functions of lexicon analysis, syntactic analysis, grammar analysis, lexicon, idiom dictionary, semantic analysis and target language synthesis. However, there is still not an applicable system available yet, since machine translation is a frontier application science relating to linguistics, computational linguistics, computer science and many other subjects. Under this situation, people switched their focus to the electronic dictionary which performs the translation at the word level and is more practical. Electronic dictionary is a kind of new dictionary which stores and utilizes the data in the electronic form. An electronic dictionary is of great flexibility. It can organize the information in the ways of hypermedia and hypertext, support various retrieval methods, offer the feature of dynamic translation, and provide explanations for the retrieved word in the form of text, audio and image.

[0004] However, most existing electronic dictionaries can perform word level translation only. In case of there is an English sentence “He takes part in the activity.”, existing electronic dictionaries can give Chinese translations for the single word “take”, while the user can not get the Chinese meaning of “take part in”.

[0005] There are some electronic dictionaries which are capable of idiom level translation. However, they have following restrictions to users:

[0006] They can give the translation when the user inputs the idiom as “take part in”. In case of an user inputs a phrase “take an active part in”, these electronic dictionaries will not be able to identify the idiom “take part in” and therefore can not give a useful Chinese translation.

[0007] If an user does not provide the correct head word in the idiom, those dictionaries will not give the idiom level translation. In the example of “take part in”, if the user looks up the word “part” or “in” in the dictionary, he can not get the translation for the idiom “take part in”.

[0008] The electronic dictionary with the idiom processing function according to the present invention will resolve the above problems.

[0009] In one illustrative aspect of the invention, an electronic dictionary with an idiom processing feature is operative to perform a technique which comprises:

[0010] storing a plurality of entries in the form of electronic data in a memory, each entry composed of a header of a first language and translated expressions in a second language corresponding to the header, wherein said headers include idioms of the first language;

[0011] registering a new entry into the memory; and

[0012] retrieving the entry corresponding to a query word from the memory;

[0013] wherein said retrieving operation includes an idiom processing operation which automatically identifies idioms included in a present sentence from text of the first language and retrieves the corresponding translated expressions in the second language.

[0014] The goal, features and advantages of this invention will be understood more fully from the detailed description given hereinbelow, taken in conjunction with the accompanying drawings.

[0015] FIG. 1 is an overview illustration of the electronic dictionary according to the present invention.

[0016] FIG. 2 illustrates the work flow for the idiom processing means for the electronic dictionary.

[0017] FIG. 3 illustrates the work flow for the transfer lexicon matching.

[0018] The electronic dictionary according to the present invention is illustrated in FIG. 1. It is to be appreciated that the “means” elements in FIG. 1 may be implemented in a computer or other processor-based device. For example, the input means 100 may be a keyboard, the storage means 200 may be memory associated with a processor, and the output means 500 may be a display. However, other conventional computer elements may be employed. Further, a processor may be employed to implement the entry registering means 300 and entry retrieving means 400 (and its constituent elements).

[0019] The input means 100 is used to input word string or text in the source language; the entry storage means 200 is used to store a plurality of entries in the form of electronic data; the entry registering means 300 is used to register new entries into the entry storage means 200; the entry retrieving means 400 is used to retrieve the entries matched with the query word from the entry storage means 200; output means 500 is used to output the entries that match the query word.

[0020] As illustrated in FIG. 1, the entry retrieving means 400 in the electronic dictionary according to this invention comprises an idiom processing means 600. The idiom processing means 600 is composed of a text capture unit 601, a sentence segmentation unit 602, a local grammar parsing unit 603 and a transfer lexicon matching unit 604.

[0021] FIG. 2 illustrates in detail the composition of idiom processing means 600 and its work flow. As illustrated in FIG. 2, when the idiom processing means 600 is activated, text capture unit 601 will capture a piece of text from screen display containing the query word, while the query word is the one pointed by the user with the cursor. Then, the sentence segmentation unit 602 will identify the sentence containing the query word. The local grammar parsing unit 603 will identify all of possible grammatical components from the sentence. The transfer lexicon matching unit 604 will match the parsing result with the entries in the lexicon, each matched entry can give a translation for a segment in the sentence. Those matched entries which cover the query word forms an entry set. After ranking the entries in the above entry set according to degree of match between the entry and the sentence, the output will be displayed on the screen.

[0022] For the English to Chinese electronic dictionary, local grammar parsing unit 603 can identify following grammatical components:

[0023] 1. noun phrase,

[0024] 2. verb phrase,

[0025] 3. preposition phrase,

[0026] 4. adjective phrase,

[0027] 5. adverb phrase,

[0028] 6. single particle.

[0029] A noun phrase has one of the following basic constructions or a composition of several basic constructions:

[0030] 1.1. single noun,

[0031] 1.2. noun+noun,

[0032] 1.3. noun+of +noun,

[0033] 1.4. noun's noun,

[0034] 1.5. adjective+noun,

[0035] 1.6. article+noun,

[0036] 1.7. pronoun+noun,

[0037] 1.8. numeral+noun.

[0038] A verb phrase has one of the following basic constructions or a composition of several basic constructions:

[0039] 2.1. single verb,

[0040] 2.2. be+verb,

[0041] 2.3. have+verb,

[0042] 2.4. adverb+verb,

[0043] 2.5. verb+adverb.

[0044] A preposition phrase has the construction of:

[0045] 3.1. preposition+noun phrase.

[0046] An adjective phrase has one of the following basic constructions:

[0047] 4.1. single adjective,

[0048] 4.2. more+adjective,

[0049] 4.3. the most+adjective.

[0050] An adverb phrase has one of the following basic constructions:

[0051] 5.1. single adverb,

[0052] 5.2. more+adverb,

[0053] 5.3. the most+adverb.

[0054] In the following, the sentence “He takes an active part in the activity.” is taken as an example to illustrate local parsing procedures:

[0055] Step I

[0056] he ->noun

[0057] take->verb

[0058] a->article

[0059] active->adjective

[0060] part->noun

[0061] part->adjective

[0062] in->preposition

[0063] in->particle

[0064] the->article

[0065] activity->noun

[0066] Step II

[0067] he: noun

[0068] take: verb

[0069] a: article

[0070] active: adjective

[0071] part: noun

[0072] part: adjective

[0073] in: preposition

[0074] in: particle

[0075] the: article

[0076] activity: noun

[0077] active part: adjective+noun->noun

[0078] the activity: article+noun->noun

[0079] Step III

[0080] he: noun

[0081] take: verb

[0082] a: article

[0083] active: adjective

[0084] part: noun

[0085] part: adjective

[0086] in: preposition

[0087] in: particle

[0088] the: article

[0089] activity: noun

[0090] active part: noun

[0091] the activity: noun

[0092] an active part: article+noun->noun

[0093] in the activity: preposition+noun->preposition phrase

[0094] While identifying grammatical components in a sentence, there may happen that one word belongs to more than one components which are of totally different part of speech. In the above sentence, the word “part” can be either an adjective or a noun, and the word “in” can be either a preposition or a particle. In this case, the result of local parsing will keep all of the possible grammatical components.

[0095] After the local grammar parsing is completed, idiom processing means 600 will activate the transfer lexicon matching unit 604 to perform matching between the entries in the transfer lexicon and the result of the local grammar parsing.

[0096] The following presents the structure of the transfer lexicon. One entry in the lexicon consists of three parts: head, grammar and translation.

[0097] The notations in the above table are interpreted as:

[0098] The head of an entry may consists more than one words, for example:

[0099] make sure<v comp(p about of)<objprep it takes the first word as the head of the entry.

[0100] In the grammar part of the entry, the first component is the part of speech of the head word.

[0101] Following the part of speech, it comes a series of grammar components. A grammar component is composed of its name and its description. It may consists of several sub-components.

[0102] In the example,

[0103] obj n fin wh it describes an object grammar component. This object has to be a noun, a finite verb phrase or a clause headed by wh.

[0104] In the example:

[0105] p about of it describes a preposition phrase. This preposition phrase has to be headed by ‘about’ or ‘of’.

[0106] In the translation part, there are grammar parts together with Chinese words as the translation of the English part of an entry.

[0107] FIG. 3 illustrates the procedures to match the entries in the transfer lexicon with the parsing result of the sentence. The followings are two examples of the transfer lexicon entry.

[0108] The first one is

[0109] take<v (Obj part) (comp (p in))<t(objprep)

[0110] The skeleton idiom for this entry is “take part in” (A skeleton idiom is an idiom which contains the necessary words only). Within the entry, “take” is the head word. “v” tells this is a verb phrase. “(Obj part)” indicates that the verb “take” needs to have the word “part” as its object. “comp” specifies a complement component after the object. “(p in)” denotes that the complement should a preposition phrase headed by “in”. “t(objprep)” shows how to translate the idiom into Chinese. “objprep” represents the object of preposition “in”.

[0111] The second example is

[0112] take<v (obj n fin wh) (comp (pt in))<t(obj)

[0113] The skeleton idiom for this entry is “take in”. Within the entry, “(obj n fin wh)” indicates that verb “take” needs to have an object. The object can be a noun phrase, a finite verb phrase or a clause leading by “what”, “where”, “when” etc. “(pt in)” denotes that the complement should be the particle “in”.

[0114] Comparing the above two entries with the parsing result of “He takes an active part in the activity.”, it gives following piece by piece matches

[0115] 1. takes vs. take<v

[0116] 2. an active part vs. (Obj part)

[0117] 3. in the activity vs. (comp (p in)) and

[0118] 1. takes vs. take<v

[0119] 2. an active part vs. (obj n)

[0120] 3. in vs. (comp (pt in))

[0121] In the above examples, both the two entries match with the sentence. If the query word is “take” or “in” both of them are in the set of dictionary lookup. However, if the query word is “part”, only the entry “take part in” is in the set of dictionary lookup.

[0122] Finally, the matched entries are ranked according to the degree of match between the entry and the parsing result. For example, take the sentence “He takes an active part in the activity”, and the query word “take”, there are three matched entries available: “take part in”, “take in” and “take”. According to the length of the skeleton idioms, i.e. number of words in the skeleton idiom, the output entries are ranked as:

[0123] 1. take part in:

[0124] 2. take in:

[0125] 3. take:

[0126] From above content, the electronic dictionary described by this invention can identify all of the idioms from the input text that contain the query word. This invention realized the idiom level intelligent translation.

[0127] For those skilled in the art, this invention allows various modifications and deformations under the condition that the spirit and the category of this invention remain unchanged; this invention intends to protect all of these modification and deformation.