KR20210068790A - Sign language interpretation system - Google Patents

Abstract

The present invention relates to a sign language interpretation system based on artificial intelligence (AI). The system comprises: a dictionary storage unit which defines a prototype dictionary that defines words derived from the same word as a single word, a thesaurus that defines words with the same or similar meaning as a single word, and defines stopwords that are not used in sign language translation among morphologically analyzed sentences, a set of stopword processing dictionary, a homonym dictionary in which different identification information is set for each homonym is stored; a morpheme analyzer for classifying morphemes with respect to the input sentence; a sign language sentence generator for generating a sign language sentence to be translated by comparing it with the dictionaries for each of the divided morphemes; a motion data extraction unit for extracting from the storage unit each motion data indicated by a sign language word code matching each morpheme constituting the generated sign language sentence; and an avatar motion display unit configured to display and control the motion of the sign language transmission avatar on the display unit according to the extracted motion data.

Description

Sign Language Interpretation System {SIGN LANGUAGE INTERPRETATION SYSTEM}

The present invention relates to a sign language interpretation system, and more particularly, to an artificial intelligence (AI)-based sign language interpretation system.

In accordance with the enforcement of the Anti-Discrimination Act of the Disabled by the government, various information service support systems for the disabled have been established and supported. For example, a significant number of hearing-impaired people do not understand Korean, but closed captioning for the disabled is a typical example.

Ordinary people think that hearing-impaired people cannot hear spoken language (Korean), but because they can communicate by reading, there will be no inconvenience in accessing or communicating with various information. However, hearing impaired people have difficulty in acquiring the spoken language (Korean) because of their hearing difficulties, and therefore have difficulties in acquiring the characters (Hangul) corresponding to the spoken language. In other words, with the exception of a few, the hearing impaired generally have difficulties in Korean literacy, limiting their ability to acquire information quickly and accurately through any medium.

In addition, the existing sign language service is also called 'Korean correspondence sign language (used when deaf people who are not proficient in sign language grammar because Korean is their first language and then start to acquire sign language try to understand the meaning of Korean in sign language) It is different from 'Korean sign language (used when a deaf person who is proficient in sign language grammar by acquiring sign language as a first language tries to understand the meaning of Korean)' used by the deaf as '. It is necessary to provide services centering on the Korean sign language that is being used.

Republic of Korea Patent Publication No. 10-1612873

Accordingly, the present invention is an invention devised to solve the above-mentioned necessity and the above-mentioned problems, and the main object of the present invention is to provide an artificial intelligence-based sign language interpretation system for strengthening media accessibility of the hearing impaired and improving civil service services.

Furthermore, another object of the present invention is to provide a sign language interpretation system based on Korean sign language for the hearing impaired,

Furthermore, the present invention aims to provide an AI-based sign language interpretation system to lay the foundation for transition from the existing closed captioning-centered information service system for the hearing impaired to a system for improving access to information services for the hearing impaired. do.

Another object of the present invention is to provide an AI-based sign language interpretation system capable of interpreting closed caption broadcasting for the hearing impaired into sign language based on Korean sign language.

In addition, the present invention is to provide an AI-based sign language interpretation system that can service web contents in sign language as well as civil service contents of government agencies, local governments, and public institutions.

Finally, another object of the present invention is to provide an AI-based sign language interpretation system centered on Korean sign language that can translate and provide the most suitable sign language to the hearing impaired through semantic analysis of Korean sentences.

Sign language interpretation system according to an embodiment of the present invention for solving the above technical problem,

Defines a prototype dictionary that defines words derived from the same word as a single word, a thesaurus that defines words with the same or similar meaning as a single word, and defines stopwords that are not used in sign language translation among morphologically analyzed sentences a dictionary storage unit for storing the stopped word processing dictionary and the homonym dictionary in which different identification information is set for each homonym;

a morpheme analyzer for classifying morphemes with respect to the input sentence;

a sign language sentence generator for generating a sign language sentence to be translated by comparing it with the dictionaries for each of the divided morphemes;

a motion data extraction unit for extracting each motion data indicated by a sign language word code matching each morpheme constituting the generated sign language sentence from a sign language dictionary;

and an avatar motion display unit configured to display and control the motion of the sign language transmission avatar on the display unit according to the extracted motion data,

Another feature of the sign language interpretation system described above is that it further includes a voice recognition text converter for converting a voice signal input through a microphone into a sentence and transmitting it to the morpheme analyzer.

In the sign language interpretation systems having the above configuration, the input sentence may be a subtitle broadcasting sentence that has been decoded and separated from a broadcast signal,

The input sentence may be a sentence requested for sign language interpretation from an external computer system through a communication network.

Furthermore, in the sign language interpretation systems of the above configuration, the sign language sentence generator,

When one of the divided morphemes is a homonym, the preceding and following words can be recognized and one of the morphemes can be selected and included in the sign language sentence,

The dictionary storage unit further stores an individual recognition dictionary in which processing rules for proper nouns such as names and places are defined. In this case, the sign language sentence generator compares the original dictionary for each divided morpheme, processes synonyms, and recognizes entities. Another feature is that the sign language sentence is generated by sequentially performing the review and homonym processing.

A sign language interpretation system according to another embodiment of the present invention is a sign language interpretation system executable in a server,

Defines a prototype dictionary that defines words derived from the same word as a single word, a thesaurus that defines words with the same or similar meaning as a single word, and defines stopwords that are not used in sign language translation among morphologically analyzed sentences a dictionary storage unit for storing the stopped word processing dictionary and the homonym dictionary in which different identification information is set for each homonym;

a morpheme analyzer for classifying morphemes for a sign language interpretation desired sentence specified on the website by the website visitor;

a sign language sentence generator for generating a sign language sentence to be translated by comparing it with the dictionaries for each of the divided morphemes;

a motion data extraction unit for extracting each motion data indicated by a sign language word code matching each morpheme constituting the generated sign language sentence from a sign language dictionary;

and an avatar motion display unit configured to display and control the motion of the sign language transmission avatar on the website according to the extracted motion data.

According to the above-described problem solving means, the present invention has the effect of providing an interpretation of the most suitable Korean sign language-based sign language for the hearing impaired through semantic analysis of Korean sentences,

In addition, the present invention interprets the sentences posted on the website into sign language based on Korean sign language and expresses them as motions of the digital elements and the non-finger elements through the sign language delivery avatar. It also provides the effect of improving

Furthermore, since the present invention can be applied to a broadcast signal receiving device such as a set-top box, there is an effect of delivering caption contents to the hearing impaired through a sign language delivery avatar for closed caption broadcasting as well as real-time closed caption broadcasting.

1 is an exemplary configuration diagram of a sign language interpretation system according to an embodiment of the present invention.
FIG. 2 is an exemplary operation flow diagram of the sign language interpretation system shown in FIG. 1. FIG.
3 is an exemplary view of a sign language dictionary according to an embodiment of the present invention.
4 is an exemplary system configuration to which the sign language interpretation system shown in FIG. 1 is applicable.
Figure 5 is another applicable system configuration diagram of the sign language interpretation system shown in Figure 1;
6 is an exemplary configuration diagram of a sign language interpretation system that can be built in the public institution server 400 shown in FIG. 5 .
7 is an exemplary operation flow diagram of the sign language interpretation system shown in FIG.

Specific structural or functional descriptions of the embodiments according to the concept of the present invention disclosed in this specification are only exemplified for the purpose of explaining the embodiments according to the concept of the present invention, and the embodiments according to the concept of the present invention are It may be implemented in various forms and is not limited to the embodiments described herein.

In addition, since the embodiments according to the concept of the present invention may have various changes and may have various forms, the embodiments will be illustrated in the drawings and described in detail in the present specification. However, this is not intended to limit the embodiments according to the concept of the present invention to specific disclosed forms, and includes all modifications, equivalents, or substitutes included in the spirit and scope of the present invention.

Hereinafter, embodiments of the present invention will be described in more detail with reference to the accompanying drawings.

First, FIG. 1 illustrates a configuration diagram of a sign language interpretation system 100 according to an embodiment of the present invention. A sign language interpretation system 100 according to an embodiment of the present invention is a collection of control program data executable by a microprocessor constituting a computer system, and as shown in FIG. 1 , a morpheme analyzer 110 and a sign language sentence generator 115 ), the motion data extraction unit 120 and the avatar motion display unit 125 may be modularized into a plurality of functional blocks.

More specifically, the sign language interpretation system 100 includes a morpheme analyzer 110 for classifying morphemes with respect to an input sentence, and a sign language sentence generator for generating a sign language sentence to be translated by comparing the divided morphemes with dictionaries to be described later ( 115), a motion data extraction unit 120 that extracts each motion data indicated by a sign language word code matching each morpheme constituting the generated sign language sentence from a sign language dictionary, and a sign language delivery avatar according to the extracted motion data. and an avatar motion display unit 125 for displaying and controlling the motion on the display unit.

Furthermore, the sign language interpretation system 100 according to the embodiment of the present invention further includes a dictionary storage unit 130 for storing various dictionary data necessary for sign language interpretation.

The dictionary is a circular dictionary in which words derived from the same word are defined as one word as shown in FIG. 1;

A thesaurus that defines words with the same or similar meaning as a single word;

Stopword processing dictionary that defines stopwords that are not used in sign language translation among morpheme-analyzed sentences;

It refers to a homonym dictionary in which different identification information is set for each homonym, and in some cases, further includes an individual recognition dictionary defining processing rules for proper nouns such as names and places.

For reference, in the original dictionary, for example, “..do, ..do, ..do, ..do” all mean words derived from the word “do”, so they can be combined into one single word like “do”. A dictionary that defines words

Thesaurus is a dictionary in which words with similar meanings, such as "blue, bluish, blue", are defined as one word, "blue."

The stopword processing dictionary is a dictionary that defines terms that are not used in sign language translation among morpheme-separated sentences.

The homonym dictionary describes words that have the same sound, such as 'fall', but have different meanings (should be expressed in a different sign language), such as "fall from the test" and "fall from the roof". and) a dictionary constructed by matching different identification information such as morpheme descriptions. A word corresponding to a homophone can be interpreted as a word by recognizing the words before and after it.

For reference, when one divided morpheme is a homonym, the sign language sentence generator 115 recognizes the preceding and following words based on AI, selects one morpheme among the homonyms, and includes the morpheme in the sign language sentence, but the identification information of the selected morpheme ( It is desirable to include parts of speech) together so that it can be used to search for sign language word codes, which will be described later.

In addition, the sign language sentence generator 115 sequentially performs archetype dictionary comparison, synonym processing, individual recognition review, and homonym processing for each divided morpheme to generate a sign language sentence, which will be described later.

For reference, the input sentence input to the morpheme analyzer 110 may be a sentence specified or input by the user of the system 100 through the UI, and when the system 100 is mounted and operated in the broadcasting signal receiver, the received broadcasting signal It may be a sentence of closed caption broadcasting separated from and decoded, or it may be a sign language interpretation request sentence transmitted from another computer or server system that can be linked through a communication network.

Meanwhile, in addition to the above-described configuration, the sign language interpretation system 100 according to an embodiment of the present invention includes a voice recognition character converter for converting a voice signal input through a microphone into a sentence and transmitting it to the morpheme analyzer 110 , and sign language. It may further include a storage unit for storing motion data indicated by the word code. The motion data indicated by the sign language word code may be matched with the sign language word code and stored in the sign language dictionary according to an implementation method.

Hereinafter, the operation of the sign language interpretation system 100 having the above-described configuration will be described in detail with reference to FIGS. 1 to 3 .

2 illustrates an operation flow of the sign language interpretation system 100 shown in FIG. 1, and FIG. 3 illustrates a sign language dictionary according to an embodiment of the present invention. The sign language dictionary shown in FIG. 3 is a delimiter of 'morpheme' constituting a sign language sentence, 'part of speech' of the morpheme, 'sign language word code' used to identify the morpheme and extract motion data, and each 'description of morpheme'. includes In some cases, it may further include a 'motion data' delimiter.

2, the morpheme analyzer 110 of the sign language interpretation system 100 according to the embodiment of the present invention is transmitted from the user of the system 100 through the UI or from another computer or server system that can be linked through the communication network. For the input sentence (step S10), morphemes are divided (step S20).

For reference, the Korean morpheme analyzer can be designed to understand the meaning of Korean sentences by using the bidirectional LSTM (Long Short-Term Memory) type morpheme analyzer to understand the meaning of the Korean sentence and select an optimal word accordingly. Such a morpheme analyzer can recognize the basic form of each morpheme constituting a word, and can remove investigations that are not necessary for sign language interpretation.

When the research and the like are removed by the morpheme analyzer 110 and the basic form of each morpheme is recognized separately, the sign language sentence generator 115 compares the divided morphemes with the dictionaries stored in the dictionary storage 130 to create a sign language sentence to be translated. generated (step S30).

Sign language sentences to be translated are generated by sequentially applying original dictionary comparison, synonym processing, individual recognition review (optional), and homonym processing for each divided morpheme.

For example, if the input sentence is “I am going to school”, a sign language sentence “I + school + going” is created, and if the sentence is “more than 100 people applied for the contest”, “contest + apply + person + It generates a sign language sentence of "one hundred + more", and if the input sentence is "Are you at your father's house?", a sign language sentence of "you + father + house + are you +?" is generated.

When the sign language sentence is generated, the motion data extraction unit 120 stores each of the motion data indicated by the sign language word code matching each morpheme constituting the generated sign language sentence, or the motion data from the sign language dictionary that matches the motion data. Extraction (step S40).

If the morpheme constituting the sign language sentence is "a" corresponding to the homonym as shown in FIG. 3, the motion data extraction unit 120 retrieves identification information of the morpheme included in the sign language sentence, for example, "part of speech" information. It can be used to identify morphemes and extract the motion data indicated by the corresponding sign language word code.

Motion data is motion-captured data for all morphemes that can be expressed by combining not only the numerical elements such as tree shape, water level, manual direction, and hand direction, but also non-digital elements such as the user's (avatar) facial expression and upper body direction and movement. It is used as data to control the positive and non-financial elements of the delivery avatar.

When the motion data is extracted, the avatar motion display unit 125 displays and controls the motion of the sign language transmission avatar on the display unit according to the extracted motion data (step S50), so that the deaf person transmits the sign language for the sign language sentence as the motion of the avatar. received, so that the meaning of the input sentence can be normally understood.

Therefore, the present invention has the effect of providing an interpretation of the most suitable Korean sign language-based sign language for the hearing impaired through semantic analysis of Korean sentences.

In the above embodiment, an embodiment of interpreting a Korean sign language-based sign language for a sentence input (cloud service-based) transmitted from a user of the system 100 through the UI or from another computer or server system that can be linked through a communication network is described. However, by applying the sign language interpretation system of the present invention to a broadcast signal receiving device such as a set-top box, sign language interpretation for closed captions as well as real-time closed captions can be performed. To explain this further with reference to 4,

First, FIG. 4 is a broadcast signal receiving apparatus to which the sign language interpretation system 100 shown in FIG. 1 is applicable, and more specifically, illustrates a partial configuration diagram of a set-top box.

Referring to FIG. 4 , the broadcast signal separation and decoding processing unit 220 separates the broadcast signal of the channel selected through the tuner 210 into video data, audio data, and caption data, and decodes the video signal, audio signal, and caption signal. handle

More specifically, in the separation of the broadcast signal, the video, audio, and subtitle signals are separated by the Demux, and each of the separated video, audio, and subtitle (corresponding to additional data) data is imaged by the video decoder, audio decoder, and data decoder, respectively. It is decoded into a signal, audio signal, and subtitle signal.

The decoded caption signal (corresponding to the sentence input in FIG. 2 ) is transmitted to the sign language interpretation system 100 , is converted into a motion of the sign language delivery avatar by the process described in FIG. 2 , and transmitted to the signal processing unit 230 .

The signal processing unit 230 synthesizes the decoded image signal and the motion image of the sign language delivery avatar, synchronizes it with the audio signal, and outputs it. The signal processing unit 230 may be implemented as an AV processor that is a one-chip IC.

Meanwhile, the control unit 240 transmits the selected channel information to the closed captioning program management server to inquire whether the caption is real-time captioning broadcast that is delayed compared to the video, and if the query result is real-time captioning broadcast, the decoded video and audio After delaying the signal for a certain period of time (to match the delay output time of subtitles in real-time closed caption broadcasting), the video signal is synthesized with the motion image of the sign language delivered avatar, and the signal processing unit 230 is controlled so that the signal is output in synchronization with the audio signal.

As a deformable method, the control unit 250 delays the decoded video and audio signals with a preset time delay value according to an implementation method, for example, any one time value within a period of 2 to 6 seconds to deliver sign language. The signal processing unit 230 may be controlled to be output together with the motion image of the avatar.

If the closed caption is not real-time caption broadcasting, simply input the closed caption and follow the same process as in FIG. 2 to generate a motion video of the sign language delivery avatar and display it together with the video.

As described above, the sign language interpretation system according to an embodiment of the present invention is applied to a broadcast signal receiving device such as a set-top box that receives and displays closed caption broadcasting as well as real-time closed caption broadcasting, thereby providing an interpretation of Korean sign language-based sign language. there is

Hereinafter, another AI-based sign language interpretation system that can translate web content as well as civil service content from government agencies, local governments, and public institutions to sign language will be explained further.

FIG. 5 exemplifies the configuration of another applicable system of the sign language interpretation system shown in FIG. 1, and the sign language interpretation system according to an embodiment of the present invention can be built, for example, in the public institution server 400 system. Therefore, in the following, reference number 400 will be commonly used as a reference number for a public institution server and a sign language interpretation system. In FIG. 5 , various types of terminals 300 for the hearing impaired are terminals that can be connected to the public institution server 400 through a communication network, and the public institution server 400 is the terminal 300 for the hearing impaired that requested sign language interpretation. It provides a sign language interpretation service.

Meanwhile, FIG. 6 exemplifies the configuration of a sign language interpretation system that can be built in the public institution server 400 shown in FIG. 5 , and the sign language interpretation system 400 shown in FIG.

Defines a prototype dictionary that defines words derived from the same word as a single word, a thesaurus that defines words with the same or similar meaning as a single word, and defines stopwords that are not used in sign language translation among morphologically analyzed sentences A dictionary storage unit 430 for storing a disambiguation dictionary, a homonym dictionary in which different identification information is set for each homonym;

a morpheme analyzer 410 for analyzing and classifying morphemes for a sign language interpretation desired sentence specified on the website by a website visitor of a public institution;

a sign language sentence generator 415 for generating a sign language sentence to be translated by comparing it with the dictionaries for each of the divided morphemes;

A motion data extraction unit 420 for extracting each motion data indicated by a sign language word code matching each morpheme constituting the generated sign language sentence from a storage unit or a sign language dictionary;

and an avatar motion display unit 425 for displaying and controlling the motion of the sign language transmission avatar on the website according to the extracted motion data.

Also in the sign language interpretation system 400 having the above configuration, the sign language sentence generator 415 is

Another feature is that, when one divided morpheme is a homophone, it recognizes the words before and after and selects one morpheme among the homophones and includes it in the sign language sentence,

Furthermore, another feature of the sign language sentence generator 415 is to include identification information (eg, part-of-speech) of the selected morpheme in the sign language sentence and use it to search for a sign language word code.

Meanwhile, the dictionary storage unit 430 further stores an individual recognition dictionary in which processing rules for proper nouns such as names and place names are defined. In this case, the sign language sentence generator 415 compares the original dictionary for each of the divided morphemes. , synonym processing, individual recognition review, and homonym processing are sequentially performed to generate the sign language sentence.

Referring to FIG. 7 showing the operation flow of the sign language interpretation system that can be built in the public institution server 400 below,

First, the deaf person accesses (visits) the website of the public institution server 400 through his or her terminal 300 (step S100). Next, the deaf person designates a sentence for which sign language interpretation is desired (hereinafter referred to as 'sign language interpretation hope sentence') in order to understand the contents written on the visited website (step S110).

Then, the deaf person requests sign language interpretation for the designated sign language interpretation request (step S120). In response to the sign language interpretation request, the morpheme analyzer 410 classifies morphemes for the designated sign language interpretation desired sentence (step S130).

As described above, when the morphemes are removed by the morpheme analyzer 410 and the basic form of each morpheme is recognized separately, the sign language sentence generator 415 translates each divided morpheme by comparing it with the dictionaries stored in the dictionary storage unit 430 . A sign language sentence is generated (step S140).

Sign language sentences to be translated are generated by sequentially applying archetype dictionary comparison, synonym processing, individual recognition review (optional), and homonym processing for each divided morpheme.

As in the previous example, if the desired sentence for sign language interpretation is “I am going to school”, a sign language sentence of “I + school + going” is created.

When the sign language sentence is generated, the motion data extraction unit 420 stores each of the motion data indicated by the sign language word code matching each morpheme constituting the generated sign language sentence, or the motion data from the sign language dictionary that matches the motion data. extraction (step S150).

When the motion data is extracted, the avatar motion display unit 425 generates a sign language video of the sign language delivery avatar that moves according to the extracted motion data (step S160) and plays it back (step S170), so that the hearing impaired terminal 300 A sign language video for the desired sentence for sign language interpretation is displayed (step S180). Depending on the implementation method, the sign language delivery avatar may be displayed on the deaf terminal 300 when a sign language interpretation is requested, and the motion of the sign language delivery avatar may be sequentially changed according to motion data transmitted from the public institution server 400 .

In another embodiment of the present invention described above, since the sentences designated by the deaf person on the website are interpreted and provided in Korean sign language-based sign language, the deaf person can post on a website provided by a public institution, local government, or portal site. Information can be provided in sign language based on Korean sign language.

Therefore, the present invention provides the effect of enhancing the media accessibility of the hearing impaired as well as improving the civil service.

For reference, the sign language interpretation system according to the embodiment may be mounted on a web server or a separate server capable of interworking with the web server to provide a sign language interpretation service.

The above has been described with reference to the embodiments shown in the drawings, which are merely exemplary, and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. For example, the sign language interpretation system according to an embodiment of the present invention can be said to be a sign language interpretation system applicable to various systems based on a cloud-based, website-based, or broadcast signal receiving device. Accordingly, the true technical protection scope of the present invention should be defined only by the appended claims.

Claims (12)

Defines a prototype dictionary that defines words derived from the same word as a single word, a thesaurus that defines words with the same or similar meaning as a single word, and defines stopwords that are not used in sign language translation among morphologically analyzed sentences a dictionary storage unit for storing the stop word processing dictionary and the homonym dictionary in which different identification information is set for each homonym;
a morpheme analyzer for classifying morphemes with respect to the input sentence;
a sign language sentence generator for generating a sign language sentence to be translated by comparing it with the dictionaries for each of the divided morphemes;
a motion data extraction unit for extracting from the storage unit each motion data indicated by a sign language word code matching each morpheme constituting the generated sign language sentence;
and an avatar motion display unit configured to display and control the motion of the sign language transmission avatar on the display unit according to the extracted motion data. The sign language interpretation system according to claim 1, further comprising a voice recognition text converter for converting a voice signal input through a microphone into a sentence and transmitting it to the morpheme analyzer. The sign language interpretation system according to claim 1 or 2, wherein the input sentence is a subtitle broadcasting sentence that is decoded and separated from a broadcast signal. The sign language interpretation system according to claim 1 or 2, wherein the input sentence is a sentence requested for sign language interpretation from an external computer system through a communication network. The method according to claim 1 or 2, wherein the sign language sentence generator,
Sign language interpretation system, characterized in that when the divided one morpheme is a homonym, the preceding and following words are recognized, and one morpheme from among the homonyms is selected and included in the sign language sentence. The method according to claim 5, The sign language sentence generator,
The sign language interpretation system, characterized in that the identification information of the selected morpheme is included in the sign language sentence to be used for the sign language word code search. The method according to claim 1 or 2, wherein the dictionary storage unit,
It further stores an entity recognition dictionary defining processing rules for proper nouns such as names and places, and the sign language sentence generator performs comparison of original dictionary, synonym processing, entity recognition review, and homonym processing for each divided morpheme. A sign language interpretation system, characterized in that it generates the sign language sentences by sequentially performing. The sign language delivery avatar according to claim 1 or 2, wherein if the input sentence is a sentence of closed caption broadcasting separated from the broadcast signal and decoded and the closed caption broadcasting is real-time broadcasting, the decoded video and audio signals are delayed for a certain time Sign language interpretation system, characterized in that it further comprises; a broadcast signal output control unit for controlling the synchronization output with the motion of. In the sign language interpretation system executable on the server,
Defines a prototype dictionary that defines words derived from the same word as a single word, a thesaurus that defines words with the same or similar meaning as a single word, and defines stopwords that are not used in sign language translation among morphologically analyzed sentences a dictionary storage unit for storing the stop word processing dictionary and the homonym dictionary in which different identification information is set for each homonym;
a morpheme analyzer for classifying morphemes for a sign language interpretation desired sentence specified on the website by the website visitor;
a sign language sentence generator for generating a sign language sentence to be translated by comparing it with the dictionaries for each of the divided morphemes;
a motion data extraction unit for extracting from the storage unit each motion data indicated by a sign language word code matching each morpheme constituting the generated sign language sentence;
and an avatar motion display unit configured to display and control the motion of the sign language delivery avatar on the website according to the extracted motion data. The method according to claim 9, wherein the sign language sentence generator,
Sign language interpretation system, characterized in that when the divided one morpheme is a homonym, the preceding and following words are recognized, and one morpheme from among the homonyms is selected and included in the sign language sentence. The method according to claim 10, wherein the sign language sentence generator,
The sign language interpretation system, characterized in that the identification information of the selected morpheme is included in the sign language sentence to be used for the sign language word code search. The method according to claim 9, wherein the dictionary storage unit,
It further stores an entity recognition dictionary defining processing rules for proper nouns such as names and places, and the sign language sentence generator performs comparison of original dictionary, synonym processing, entity recognition review, and homonym processing for each divided morpheme. A sign language interpretation system, characterized in that it generates the sign language sentences by sequentially performing.

Images