JP2012208639A - Pictograph application sentence generation device, method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To apply pictographs following the actual usage of pictographs.SOLUTION: A pictograph application sentence generation device is configured to extract a pictograph including the meaning of a word which matches a word Ki included in an input sentence from a dictionary for the meanings of pictographs in which a plurality of sets of pictographs and the meanings expressed by those pictographs are stored, and to, when the plurality of pictographs are extracted, select the pictograph in which a semantic distance between the context of the input sentence and the pictograph is the most closest, and to calculate a distance (t) between the concept vector of the word Ki and the concept vector of the selected pictograph as a semantic distance between the word Ki and the selected pictograph. When the distance (t) is smaller than a threshold (k), the word Ki is determined so as to be replaced with the pictograph, and when the distance (t) is larger than the threshold (k), the pictograph is determined so as to be added after the word Ki.

Description

  The present invention relates to a pictogram-added sentence generation apparatus, method, and program, and more particularly, to a pictogram-added sentence generation apparatus and method for generating a pictogram-added sentence by replacing or adding a pictogram with a word included in an input sentence. And the program.

  Conventionally, the input of pictograms has been supported in character input devices such as mobile phones and personal computers. For example, when a specific end-of-sentence string such as “is” or “mas” is entered, the end-of-sentence character is automatically inserted by automatically inserting candidate text strings such as pictographs and punctuation that are likely to follow the end-of-sentence string. There has been proposed a character input device that can save input labor (see, for example, Patent Document 1).

  In fact, when a user manually adds a pictograph to an input sentence, if the word can be estimated using only the pictograph, the word is replaced with the pictograph, and the meaning is very close to the word. However, when it is difficult to estimate a word only with pictograms, pictograms are often used in such a way that a pictogram is added to the word.

JP 2006-331066 A

  However, the technique disclosed in Patent Document 1 specializing in input support for sentence end characters has a problem that pictograms that do not take into account the contents of the text may be displayed as input candidates. There is also a problem in that the insertion position of the pictogram is only inserted at the end of the sentence and does not conform to the actual usage of the pictogram that may be used in the sentence.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide a pictogram-added sentence generation apparatus, method, and program capable of giving pictograms in accordance with actual pictogram usage.

  In order to achieve the above object, a pictographic character sentence generation device according to the present invention matches a word included in an input sentence from a pictogram meaning storage unit that stores a plurality of sets of pictograms and at least one meaning represented by the pictogram. Matching pictogram extraction means for extracting a pictogram representing meaning, semantic feature amount of the pictogram extracted by the matching pictogram extraction means, and semantic distance between the pictogram and the word obtained from the semantic feature quantity of the word Is determined to be replacement when the pictographic word distance threshold is smaller than a predetermined pictogram word distance threshold, and is determined to be added when the distance is equal to or greater than the pictographic word distance threshold. And the determination means replaces the word with the pictogram extracted by the matching pictogram extraction means, and the determination means determines to add. And generating means for generating a pictographic character sentence in which a pictographic character is added to the input sentence by adding a pictographic character extracted by the matching pictographic character extracting means before or after the word. Has been.

  According to the pictographic character sentence generation device of the present invention, the pictogram meaning storage unit stores a plurality of sets of pictograms and at least one meaning represented by the pictograms. Here, the pictogram is a symbol in which the meaning of the meaning is illustrated, and is a symbol that can be input as a character, that is, an illustrated symbol character. The matching pictogram extracting means extracts a pictogram representing the meaning that matches the word included in the input sentence from the pictogram meaning storage means. When the semantic distance between the pictogram and the word obtained from the semantic feature quantity of the pictogram extracted by the matching pictogram extraction section and the word semantic feature quantity is smaller than a predetermined pictogram word distance threshold, Decides to replace the pictogram giving method, and if the semantic distance between the pictogram and the word is equal to or greater than the pictogram word distance threshold, decides to add the pictogram giving method. When the generation means is determined to be replaced by the determining means, the word is replaced with a pictogram extracted by the matching pictogram extracting means, and when it is determined to be added by the determining means, before or after the word By adding the pictogram extracted by the matching pictogram extracting means, a pictogram giving sentence in which a pictogram is given to the input sentence is generated.

  As described above, when the semantic distance between the word and the pictogram is smaller than the threshold, the word and the pictogram are replaced. When the semantic distance between the word and the pictogram is equal to or larger than the threshold, the pictogram is added to the word. In order to use different emoticon assignment methods, if a word can be expressed only with a pictogram, the word is replaced with a pictogram, and if it is difficult to estimate the word with only the pictogram, a pictogram is added to the word. It is possible to add pictograms in accordance with the actual usage of pictograms.

  Further, the pictographic character sentence generation device of the present invention is stored in the pictogram word meaning storage means when a pictogram representing a meaning that matches the word included in the input sentence is not extracted by the matching pictogram extraction means. A pictogram representing a meaning of a semantic distance between each word meaning and the word obtained from the semantic feature quantity of each word meaning and the word is smaller than a predetermined word meaning word distance threshold. A short-distance pictogram extraction unit that extracts from the storage unit may be included, and the generation unit may add the pictogram extracted by the short-distance pictogram extraction unit before or after the word. As a result, even if there is no pictogram that matches the meaning of the word and the pictogram, a pictogram with a close semantic distance between the word and the pictogram is added to the word. The emoji can be given in accordance with the actual usage of the pictogram.

  Further, the pictogram-added sentence generation apparatus of the present invention provides a pictogram meaning sentence stored in the pictogram meaning storage means and a pictorial character obtained from the feature quantity indicating the context of the input sentence and the input sentence. A pictograph whose semantic distance is smaller than a predetermined pictogram context distance threshold value can be configured to include a context pictogram extracting means for extracting from the pictogram meaning storage means. The pictogram extracted by the context pictogram extracting means can be added at the end. This gives pictograms that follow the actual usage of pictograms by adding pictograms that reflect the meaning of the whole sentence, so that pictograms that have a close semantic distance between the input text and pictograms are given to the beginning or end of the text. Can be given.

  Further, the pictogram giving sentence generation method of the present invention is a pictogram giving sentence generating method in a pictogram giving sentence generating device including a matching pictogram extracting means, a determining means, and a generating means, wherein the matching pictogram extracting means is a pictogram And at least one meaning represented by the pictogram, a pictogram representing the meaning of the word that matches the word included in the input sentence is extracted from the pictogram meaning storage means that stores a plurality of pairs, and the determining means includes the matching pictogram extraction means When the semantic distance between the extracted pictogram and the word obtained from the semantic feature of the word is smaller than a predetermined pictogram word distance threshold, the pictogram is added. The method is determined to be replaced, and when the distance is equal to or greater than the pictogram word distance threshold, the pictogram assigning method is determined to be added, and the generating means is replaced by the determining means. If determined, the word is replaced with the pictogram extracted by the matching pictogram extracting means, and if determined to be added by the determining means, extracted by the matching pictogram extracting means before or after the word This is a method of generating a pictographic character sentence by adding a pictograph to the input sentence by adding the pictograph.

  Further, the pictogram-added sentence generation method of the present invention is a pictogram-added sentence generation method in a pictogram-added sentence generation device further including a short-distance pictogram extraction means, wherein the short-distance pictogram extraction means includes a word included in the input sentence When the pictogram representing the meaning of the word is not extracted by the matching pictogram extraction means, the semantic feature quantity of each meaning stored in the pictogram meaning storage means and the semantic feature quantity of the word A pictogram representing a sense of meaning that a semantic distance between each sense of meaning and the word is smaller than a predetermined meaning word distance threshold is extracted from the pictogram meaning storage means, and the generation means is configured to extract the proximity before or after the word. This is a method of adding pictograms extracted by the distance pictogram extracting means.

  Further, the pictogram giving sentence generating method of the present invention is a pictogram giving sentence generating method in a pictogram giving sentence generating device further including a context pictogram extracting means, and the context pictogram extracting means is stored in the pictogram meaning storage means. A pictogram meaning a pictogram whose semantic distance between each pictogram and the input sentence obtained from the semantic feature quantity of each pictogram and a feature quantity indicating the context of the input sentence is smaller than a predetermined pictogram context distance threshold. Extracted from storage means, and the generating means adds the pictogram extracted by the context pictogram extracting means to the beginning or end of the input sentence.

  Moreover, the pictorial character provision sentence generation program of this invention is a program for functioning a computer as each means which comprises said pictogram provision sentence generation apparatus.

  As described above, according to the pictograph-added sentence generation device, method, and program of the present invention, when the semantic distance between the word and the pictogram is smaller than the threshold, the word and pictogram are replaced, If the semantic distance to the emoji is greater than or equal to the threshold, the emoji is used differently, so that the emoji is added to the word. When it is difficult to estimate a word only with pictograms, it is possible to add pictograms according to the actual usage of pictograms by adding pictograms to the words.

It is a figure which shows an example of the pictogram used by this Embodiment. It is a block diagram which shows the functional structure of the pictograph provision sentence generation apparatus of this Embodiment. It is a figure which shows an example of an emoji character dictionary. It is a figure which shows an example of an emoji word meaning dictionary. It is a flowchart which shows the content of the pictogram provision sentence production | generation routine in the pictogram provision sentence production | generation apparatus of this Embodiment. It is a flowchart which shows the content of the character pictogram replacement processing routine. It is a figure for demonstrating allocation of character ID to an input sentence. It is a figure which shows an example of the pictogram memorize | stored in the character substitution pictogram storage part. It is a flowchart which shows the content of the word matching pictogram replacement addition processing routine. It is a figure for demonstrating allocation of word ID to an input sentence. It is a figure which shows an example of the pictogram memorize | stored in the word substitution pictogram storage part. It is a figure which shows an example of the pictogram memorize | stored in the additional pictogram storage part. It is a flowchart which shows the content of the word short distance pictogram addition process routine. It is a flowchart which shows the content of a context pictogram addition process routine. It is a figure which shows an example of the pictogram memorize | stored in the sentence end pictogram storage part. It is a flowchart which shows the content of the production | generation process routine. It is a figure for demonstrating the production | generation of a pictograph provision sentence.

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

  First, the pictograph used in this embodiment is a pictograph that can be input from a mobile terminal as shown in FIG. These pictograms have the following four types depending on their usage.

(1) Pictograph used as a substitute for text (2) Pictograph used as a substitute for word (3) Pictograph used in addition to a word (4) Pictograph used as a summary at the end of a sentence

  In the case of (3), although the pictogram has a very close meaning to the word, it is used when it is difficult to estimate the word only from the pictogram. What is used at the end of the sentence in (4) is given a reflection of the meaning of the whole sentence. The pictograph-added sentence generation device of the present embodiment automatically gives pictographs to an input sentence in accordance with the usage of these pictographs.

  The pictograph-added sentence generation device 10 according to the present embodiment includes a CPU (Central Processing Unit), a RAM (Random Access Memory), and a ROM (Read that stores a program for executing a pictograph-added sentence generation processing routine described later. Only memory). As shown in FIG. 2, this computer functionally includes a morphological analysis unit 12, a character pictogram replacement unit 14, a word matching pictogram replacement addition unit 16, a word short-distance pictogram addition unit 18, a context pictogram addition unit 20, a character It can be expressed by a configuration including a substitution pictogram storage unit 22, a word substitution pictogram storage unit 24, an additional pictogram storage unit 26, a sentence end pictogram storage unit 28, and a pictograph giving sentence generation unit 30.

  The morpheme analyzer 12 performs a morpheme analysis on the input sentence inputted by a known morpheme analyzer and outputs an analysis result.

  The character pictogram replacement unit 14 performs processing corresponding to the usage of the pictogram (1) described above, determines whether one character included in the input sentence can be replaced with a pictogram, and converts the replaceable character into a pictogram. Replace. For example, as shown in FIG. 3, a pictogram character dictionary in which pictograms and characters are associated with each other is determined in advance, and the pictogram character dictionary is sequentially referred to from the first character of the input sentence, and the pictogram corresponding to the character is selected. Extract. When there is a pictogram corresponding to the character, the character replacement pictogram storage unit 22 stores the position ID indicating the position of the character to be replaced with the pictogram and the searched pictogram.

  The word matching pictogram replacement adding unit 16 performs processing corresponding to the usage of pictograms (2) and (3) above, and determines whether or not one word included in the input sentence subjected to morphological analysis can be replaced with a pictogram. Judgment is made, and replaceable characters are replaced with pictograms. For example, for all the pictograms used in the pictogram giving sentence generation apparatus 10 of the present embodiment, a pictogram word meaning dictionary associating the pictograms with the meanings represented by the pictograms as shown in FIG. 4 is determined in advance. . Then, referring to the pictogram word meaning dictionary sequentially from the first word of the input sentence, pictograms including the word as the word meaning are extracted. In the pictogram meaning dictionary, a plurality of meanings may be associated with one pictogram, or the same meaning may be associated with different pictograms. Further, since all pictograms are stored in the pictogram word meaning dictionary, there are some that overlap with the pictograms stored in the pictogram dictionary. That is, a pictogram that can be replaced with one character is stored in both the pictogram word meaning dictionary and the pictogram character dictionary.

  In addition, when a plurality of pictograms are extracted for one word, the word matching pictogram replacement adding unit 16 selects one pictogram that most closely matches the context of the input sentence.

  In addition, the word matching pictogram replacement adding unit 16 determines whether to replace the extracted pictogram with a word or to add to the word based on the semantic distance between the word and the pictogram. If the distance is close, it is determined that the word and pictogram can be replaced. If the distance is not close, it is determined that the word and pictogram cannot be replaced, and a decision is made to add the pictogram to the word. To do. When replacing a word with a pictogram, the position ID indicating the position of the word to be replaced with the pictogram and the extracted pictogram are stored in the word replacement pictogram storage unit 24. In addition, when adding a pictograph to a word, the location ID indicating the position to add the pictograph and the extracted pictograph are stored in the additional pictograph storage unit 26.

  The word short-distance pictogram adding unit 18 performs processing corresponding to the above-described pictogram usage in (3). When the word matching pictogram replacement adding unit 16 does not extract a pictogram including a word as a meaning, Based on the semantic distance between the word and the meaning of the pictogram, a pictogram close to the meaning of the word is extracted from the pictogram semantic dictionary. The pictograms extracted here are pictograms that do not include words as meanings but include meanings that are very close in semantic distance to the words. For example, when “emoticon expressing cake” and “cake” are registered in the pictogram dictionary as the meaning of the pictogram, the word “dessert” and the meaning of the cake pictogram “cake” for the word “dessert” It is thought that the semantic distance to is very close. In this way, by using the semantic distance between the word and the meaning of the pictogram, the word “dessert” in the input sentence even when the “pictogram representing the cake” does not include “dessert” as the meaning in the pictogram meaning dictionary As a pictograph to be added to the “pictograph representing the cake” is extracted.

  Further, when a pictogram close to the meaning of the word is extracted, the word near distance pictogram adding unit 18 stores the position ID indicating the position to which the pictogram is added and the extracted pictogram in the additional pictogram storage unit 26. Since the pictograms extracted by the word short distance pictogram adding unit 18 are pictograms not extracted by the word matching pictogram replacement adding unit 16, it is assumed that it is difficult to estimate the word from the pictograms. The addition method of pictograms is only addition.

  The context pictogram adding unit 20 executes processing corresponding to the pictogram usage described in (4) above, and extracts pictograms having a semantic distance close to the context of the input sentence from the pictogram word meaning dictionary. Is stored in the sentence end pictogram storage unit 28.

  The pictograph-added sentence generation unit 30 replaces the pictograph with the input sentence based on the information stored in the character substitution pictogram storage unit 22, the word substitution pictogram storage unit 24, the additional pictogram storage unit 26, and the sentence end pictogram storage unit 28. In addition, a pictographic text is generated.

  Next, with reference to FIG. 5, a pictographic character sentence generation process routine executed in the pictorial character sentence generation device 10 of the present embodiment will be described.

  In step 100, an input sentence is acquired. Here, it is assumed that the input sentence “I went out by car because it was sunny!” Was acquired. Next, in step 200, a character / pictogram replacement process is executed.

  Here, with reference to FIG. 6, a character pictogram replacement processing routine will be described.

  In step 202, for example, as shown in FIG. 7, Mp (p = 1, 2,..., N: n is the number of characters included in the input sentence) is assigned as the character ID in order from the first character of the input sentence. Also, 1 is set in the variable p.

  Next, in step 204, the character whose character ID is Mp (the p-th character in the input sentence) is compared with the pictogram character dictionary. Next, in step 206, it is determined whether or not a pictograph corresponding to the character Mp exists in the pictograph character dictionary. If it exists, the pictograph is extracted and the process proceeds to step 208, where Mp as the position ID indicating the position of the character to be replaced with the pictogram and the pictogram extracted in step 206 are used as the character replacement pictogram storage unit. 22 to store. For example, referring to the pictogram character dictionary of FIG. 3 for the character M15 = “!”, Since there is a pictogram that matches the character “!”, This pictogram is extracted. Then, as shown in FIG. 8, it is stored in the character replacement pictogram storage unit 22.

  Next, in step 210, it is determined whether or not processing has been completed for all characters of the input sentence by determining whether p has reached n. If p ≠ n, the process proceeds to step 212, p is incremented by 1, and the process returns to step 204 to repeat the process. If p = n, the process returns to the pictograph-added sentence generation process (FIG. 5) and proceeds to step 300.

  In step 300, the input sentence acquired in step 100 is subjected to morphological analysis. Next, in step 400, a word matching pictogram replacement addition process is executed.

  Here, with reference to FIG. 9, a word matching pictogram replacement processing routine will be described.

  In step 402, for example, as shown in FIG. 10, Ki (i = 1, 2,..., M: m is used as the input sentence in order from the first word using the morphological analysis result of the input sentence. Number of words included). Also, 1 is set to the variable i.

  Next, in step 404, the character with the word ID Ki (i-th word in the input sentence) is compared with the pictogram meaning dictionary. Next, in step 406, it is determined whether there is a pictograph that includes the word Ki in the meaning in the pictograph meaning dictionary. If it exists, the pictograph is extracted and the process proceeds to step 408. If it does not exist, the process proceeds to step 500 to execute a word short distance pictogram adding process to be described later.

  In step 408, it is determined whether or not there are a plurality of pictograms extracted in step 406. If there are a plurality of pictograms, the process proceeds to step 410. If there is only one, the process skips step 410 and step 412. Migrate to

In step 410, the pictogram having the closest semantic distance between the context of the input sentence and the pictogram is selected from the plurality of pictograms extracted in step 406 as the pictogram along the context of the input sentence. Specifically, first, a feature vector expressing the context of the entire input sentence is calculated. As the feature vector expressing the context, for example, using a concept vector set generated by a method disclosed in Japanese Patent Application Laid-Open No. 2007-317132, a concept vector a _{Ki of} each word in the input sentence is obtained, and these concept vectors are obtained. Is the feature vector expressing the context of the input sentence. The center of gravity of the concept vector is calculated by the following equation (1).

Further, the concept vector of each of the extracted pictograms is obtained by the same method as that for _obtaining the concept vector a _{Ki of} each word. Then, the distance between each obtained concept vector of pictographs and the feature vector expressing the context is calculated, and the pictograph with the smallest distance is selected.

Next, in step 412, the semantic distance between the word Ki and the selected pictogram is calculated. Specifically, the distance (t) between the concept vector a _Ki of the word Ki obtained in the process of step 410 and the concept vector of the selected pictogram is calculated.

  Next, in step 414, it is determined whether or not the distance (t) calculated in step 412 is smaller than the threshold value (k). This determination is to determine whether or not the word Ki can be estimated from the selected pictograph, that is, whether or not the word Ki can be replaced with a pictograph. The threshold (k) is determined in advance by a learning process or the like for this determination. If distance (t) <threshold value (k), it is determined that the word Ki can be estimated from the pictograph, that is, can be replaced, and the process proceeds to step 416. On the other hand, if distance (t) ≧ threshold value (t), the word Ki cannot be estimated from the pictograph, that is, cannot be replaced. Therefore, it is determined that the pictograph is added, and the process proceeds to step 418.

  In step 416, Ki as the position ID indicating the position of the word to be replaced with the pictogram and the pictogram extracted in step 406 or selected in step 410 are stored in the word replacement pictogram storage unit 24. For example, if it is determined that the word K4 = “car” can be replaced with a pictograph representing the car, the word K4 is stored in the word replacement pictograph storage unit 24 as shown in FIG.

  On the other hand, in step 418, Ki as the position ID indicating the position of the word to which the pictogram is added and the pictogram extracted in step 406 or selected in step 410 are stored in the additional pictogram storage unit 26. For example, when it is determined that a pictograph expressing clear is added for the word K1 = “clear”, it is stored in the additional pictogram storage unit 26 as shown in FIG.

  Next, in step 420, it is determined whether or not all words of the input sentence have been processed by determining whether i has become m. If i ≠ m, the process proceeds to step 422, i is incremented by 1, and the process returns to step 404 to repeat the process. If i = m, the process returns to the pictograph-added sentence generation process (FIG. 5) and proceeds to step 600.

  Here, with reference to FIG. 13, the word short distance pictogram adding process routine executed in step 500 of the word matching pictogram replacement adding process (FIG. 9) will be described.

In step 502, the semantic distance between the word Ki and each meaning included in the pictogram meaning dictionary is calculated. Specifically, the distance (o) between the concept vector a _Ki of the word Ki obtained in the process of step 410 and the meaning concept vector obtained by the same process is calculated.

  Next, in step 504, it is determined whether or not there is a pictograph whose distance (o) calculated in step 502 is smaller than a threshold (Y). This determination is to determine whether or not there is a pictogram that does not include the meaning of the word that matches the word, but serves to emphasize the word by being added to the word. As the threshold (Y), a value for this determination is determined in advance by a learning process or the like. If there is a pictograph of distance (o) <threshold (Y), the pictograph is extracted and the process proceeds to step 506.

  In step 506, Ki as the position ID indicating the position of the word to which the pictogram is added and the pictogram extracted in step 506 are stored in the additional pictogram storage unit 26, and word matching pictogram replacement addition processing (FIG. 9). And return to step 420. On the other hand, if it is determined in step 504 that there is no pictograph of distance (o) <threshold (Y), step 506 is skipped and the process returns.

  Next, with reference to FIG. 14, the context pictogram addition process routine executed in step 600 of the pictograph giving sentence generation process (FIG. 5) will be described.

  In step 602, a semantic distance between the context of the input sentence and each pictogram included in the pictogram meaning dictionary is calculated. Specifically, the distance (u) between the feature vector representing the context of the input sentence obtained in the process of step 410 and the concept vector of each pictogram included in the pictogram word meaning dictionary obtained by the same process as in step 410. ) Is calculated.

  Next, in step 604, it is determined whether or not there is a pictograph whose distance (u) calculated in step 602 is smaller than the threshold value (V). This determination is to determine whether or not there is a pictograph representing the entire input sentence. As the threshold (V), a value for this determination is determined in advance by a learning process or the like. If there is a pictograph whose distance (u) <threshold (V), the pictograph is extracted and the process proceeds to step 606.

  In step 606, the pictograph extracted in step 604 is stored in the sentence end pictogram storage unit 28 as shown in FIG. In addition, since the addition position of the pictogram stored in the sentence end pictogram storage unit 28 is determined in advance (here, the end of the sentence), the character substitution pictogram storage unit 22, the word substitution pictogram storage unit 24, and the additional pictogram storage unit 26 Unlike the above, the item of the position ID indicating the position where the pictogram is replaced or added is not included. Then, the process returns to the pictograph-added sentence generation process (FIG. 5) and proceeds to step 700. On the other hand, if it is determined in step 604 that there is no pictograph of distance (u) <threshold (V), step 606 is skipped and the process returns.

  Next, with reference to FIG. 16, the generation process routine executed in step 700 of the pictograph attached sentence generation process (FIG. 5) will be described.

  In step 702, the pictogram stored in the character replacement pictogram storage unit 22 is read, and the character indicated by the stored position ID is replaced with the pictogram. For example, as shown in FIG. 17A, the character “!” In the input sentence is replaced with a pictograph.

  Next, in step 704, the pictogram stored in the word replacement pictogram storage unit 24 is read. Next, in step 706, it is determined whether or not the pictograph read out in step 704 is a pictograph that has already been assigned in the replacement process of the text in step 702. In the case of an unassigned pictograph, the process proceeds to step 708, and the word and pictograph indicated by the stored position ID are replaced. For example, as shown in FIG. 17B, the word “car” in the input sentence is replaced with a pictograph. On the other hand, if the pictograph has already been assigned, step 708 is skipped and the process proceeds to step 710.

  In step 710, the pictogram stored in the additional pictogram storage unit 26 is read. Next, in step 712, it is determined whether or not the pictograph read out in step 710 is a pictograph that has already been assigned in the replacement process of the character and pictogram in step 702 or the replacement process of the word and pictogram in step 708. judge. In the case of an unassigned pictograph, the process proceeds to step 714, and the pictograph is added after the word indicated by the stored position ID. For example, as shown in FIG. 17C, a pictograph is added after the word “sunny” of the input sentence. On the other hand, if the pictograph has already been assigned, step 714 is skipped and the process proceeds to step 716.

  In step 716, the pictogram stored in the sentence end pictogram storage unit 28 is read. Next, in step 718, the pictogram read in step 716 is replaced with the character and pictogram replacement process in step 702, the word and pictogram substitution process in step 708, or the pictogram to the word in step 714. It is determined whether or not the pictograph has already been assigned in the addition process. In the case of an unassigned pictograph, the process proceeds to step 720, and the pictograph is added to the end of the input sentence. For example, as shown in FIG. 17D, a pictograph is added to the end of the sentence. On the other hand, if the pictograph has already been assigned, step 720 is skipped and the process proceeds to step 722.

  In step 722, a pictograph-added sentence generated by replacing or adding a pictograph to the input sentence is output, and the process ends.

  As described above, according to the pictographic character sentence generation device of the present embodiment, when the distance between the concept vector of the word and the concept vector of the pictograph is smaller than the threshold, the word and the pictograph are replaced, and the distance is When the threshold value is exceeded, the emoticon assignment method is used differently so that a pictogram is added to the word. If the word can be expressed only with a pictogram, the word is replaced with the pictogram, and the word is estimated only with the pictogram. When it is difficult to add pictograms in accordance with the actual usage of pictograms, it is possible to add pictograms to words.

  In addition, even when the word and the meaning of the pictogram do not match, an emoji that is close to the word concept vector and the meaning vector of the pictogram is added to the word. It is possible to give pictograms according to the actual usage of pictograms.

  In addition, in order to add a pictograph with a close distance between the feature vector expressing the context of the entire input sentence and the concept vector of the pictograph at the end of the sentence, the pictograph reflecting the meaning of the whole sentence is added. Can be added.

  Further, since substitution with pictograms is performed not only in word units and sentence units but also in character units, it is possible to give more detailed pictograms.

  In the above embodiment, when a pictograph is added to a word, the case where it is added after the word has been described. However, it may be added before the word. Moreover, although the case where a pictograph representing the entire input sentence is added at the end of the input sentence has been described, it may be added at the beginning of the input sentence. Further, the case where a pictogram representing the entire input sentence is added has been described. However, the input sentence may be decomposed for each phrase, and the pictogram representing each phrase may be added before or after each phrase.

  In the above embodiment, when determining whether to replace or add a pictogram in the word matching pictogram replacement giving process, the concept vector of the word and pictogram is obtained, and the semantic distance between the word and the pictogram is calculated. Although the case of calculating has been described, a semantic distance between a concept vector or a pictograph and a meaning may be registered in advance. For example, in the pictogram word meaning dictionary, [“pictogram (b)”: “meaning 1 (c1)”, “meaning 2 (c2)”, “meaning 3 (c3)”] (() is a concept vector) You can register. In this case, when a certain word in the input sentence matches the meaning 1 of the pictogram word meaning dictionary, the semantic distance (t) between the word and the pictogram is expressed as the concept vector b and the concept vector registered in the pictogram word meaning dictionary. It can be calculated as a distance from c1. Also, register as [[Emoji]: “Glossy 1 (t1)”, “Glossy 2 (t2)”, “Glossy 3 (t3)”] (() is the semantic distance to the glyph) It can also be left. In this case, when a certain word in the input sentence matches the meaning 1 in the pictogram meaning dictionary, the semantic distance (t) between the word and the pictogram uses (t1) registered in the pictogram meaning dictionary as it is. be able to.

  Further, in the above embodiment, the case of processing all the words included in the input sentence in the processing of the word matching pictogram replacement adding unit and the word short distance pictogram adding unit has been described, but only words that are independent words May be processed.

  The present invention is not limited to the above-described embodiment, and various modifications and applications are possible without departing from the gist of the present invention.

  In addition, the above-described pictographic character generation sentence generation apparatus has a computer system inside, but the “computer system” includes a homepage provision environment (or display environment) if a WWW system is used. Shall be.

  In the present specification, the embodiment has been described in which the program is installed in advance. However, the program can be provided by being stored in a computer-readable recording medium.

DESCRIPTION OF SYMBOLS 10 Pictogram provision sentence production | generation apparatus 12 Morphological analysis part 14 Character pictogram substitution part 16 Word matching pictogram substitution addition part 18 Word short-distance pictogram addition part 20 Context pictogram addition part 22 Character substitution pictogram storage part 24 Word substitution pictogram storage part 26 Additional pictogram memory Part 28 End-of-sentence pictogram storage part 30 Emoji-giving sentence generation part

Claims (7)

A matching pictogram extracting means for extracting a pictogram representing a meaning matching the word included in the input sentence from a pictogram meaning storage means storing a plurality of sets of pictograms and at least one meaning represented by the pictogram;
When the semantic distance between the pictogram and the word obtained from the semantic feature quantity of the pictogram extracted by the matching pictogram extracting means and the semantic feature quantity of the word is smaller than a predetermined pictogram word distance threshold. Determining the substitution of the pictogram giving method, and when the distance is greater than or equal to the pictogram word distance threshold, the decision means for determining the pictogram grant method to add,
When it is determined to be replaced by the determining means, the word is replaced with a pictogram extracted by the matching pictogram extracting means, and when it is determined to be added by the determining means, before or after the word Generating means for generating a pictographic character sentence in which a pictographic character is added to the input sentence by adding a pictographic character extracted by the matching pictographic character extracting means;
An emoji-giving sentence generation device including When a pictogram representing a meaning that matches a word included in the input sentence is not extracted by the matching pictogram extracting means, each semantic feature amount of the meaning stored in the pictogram meaning storage means, and A short distance pictogram extracting means for extracting, from the pictogram meaning storage means, a pictogram representing a sense of meaning whose semantic distance between each of the meanings obtained from the semantic feature and the word is smaller than a predetermined meaning word distance threshold; ,
The pictograph-added sentence generation apparatus according to claim 1, wherein the generation unit adds the pictograph extracted by the short-distance pictograph extraction unit before or after the word. The semantic distance between each of the pictograms and the input sentence obtained from the semantic feature quantity of each pictogram stored in the pictogram meaning storage means and the feature quantity indicating the context of the input sentence is a predetermined pictogram context distance. A contextual pictogram extracting means for extracting a pictogram smaller than a threshold from the pictogram meaning storage means;
The pictograph-added sentence generation apparatus according to claim 1, wherein the generation unit adds the pictograph extracted by the context pictograph extraction unit to the beginning or end of the input sentence. A pictogram giving sentence generating method in a pictogram giving sentence generating device including a matching pictogram extracting means, a determining means, and a generating means,
The matching pictogram extracting means extracts a pictogram representing a meaning that matches a word included in the input sentence from a pictogram meaning storing means storing a plurality of sets of pictograms and at least one meaning represented by the pictogram,
The determining means includes a predetermined pictogram word distance in which the semantic distance between the pictogram and the word obtained from the semantic feature quantity of the pictogram extracted by the matching pictogram extraction section and the semantic feature quantity of the word is determined in advance. If the threshold is smaller than the threshold, the method for giving the pictogram is determined to be replaced, and if the distance is greater than or equal to the threshold for the pictogram word distance, the method for giving the pictogram is determined to be added.
The generation unit replaces the word with a pictogram extracted by the matching pictogram extraction unit when the determination unit determines to replace, and when the generation unit determines to add, the word A pictogram-added sentence generation method for generating a pictogram-added sentence in which a pictogram is added to the input sentence by adding the pictogram extracted by the matching pictogram extracting means before or after. A pictogram giving sentence generating method in a pictogram giving sentence generating device further including a short distance pictogram extracting means,
The short-distance pictogram extracting means means the meaning of each meaning stored in the pictogram meaning storage means when the pictogram representing the meaning matching the word included in the input sentence is not extracted by the matching pictogram extracting means. A pictogram representing a meaning of a semantic distance between each word meaning obtained from the characteristic feature amount and the semantic feature amount of the word and the word is smaller than a predetermined word meaning word distance threshold value is extracted from the pictogram word meaning storage means And
The pictograph-added sentence generation method according to claim 4, wherein the generation unit adds the pictograph extracted by the short-range pictograph extraction unit before or after the word. A pictogram-added sentence generation method in a pictogram-added sentence generation device further including a context pictogram extracting means,
The contextual pictogram extracting means has a semantic distance between each of the pictograms obtained from the semantic feature quantity of each pictogram stored in the pictogram meaning storage means and a feature quantity indicating the context of the input sentence and the input sentence. , A pictogram smaller than a predetermined pictogram context distance threshold is extracted from the pictogram meaning storage means,
The pictograph-added sentence generation method according to claim 4 or 5, wherein the generation unit adds the pictograph extracted by the context pictograph extraction unit to the beginning or end of the input sentence.   A pictograph-added sentence generation program for causing a computer to function as each means constituting the pictograph-added sentence generation device according to any one of claims 1 to 3.

Images