JP5849690B2 - Character input program and information processing apparatus - Google Patents

Description

  The present invention relates to a program for causing a computer to function as a character input device for inputting characters to a running application, and an information processing device having a character input function.

  Many portable information processing devices typified by mobile phones are configured to input various pictograms in order to eliminate the monotony of character input and increase the enjoyment of the user. For example, by registering a combination of a pictogram and a reading of a concept word corresponding to the pictogram, and by calling a pictogram candidate together with a normal text notation candidate according to the input of the reading, Some have reduced such labor (see Patent Document 1).

  Further, many portable information processing devices predict word candidates (hereinafter referred to as “connection prediction candidates”) that are input subsequent to the input word / phrase when the input object word / phrase is confirmed. A function is provided for determining a connection prediction candidate that is displayed and selected by the user as an input target (see, for example, Patent Document 2). There are also models that use this function to connect pictographs representing the concept of the word / phrase that has been confirmed immediately before and are displayed as prediction candidates. According to this pictogram prediction function, a user can input a concept pictogram (for example, a pictogram representing a cherry blossom) corresponding to the phrase following a phrase (for example, “sakura”) in normal text notation. , Can enhance the interest of character input.

JP 2000-148748 A JP 2008-293403 A

  In order to extract a pictographic candidate as a connection prediction candidate, it is necessary to register various pictographs in association with words or phrases that match the concept of the picture. However, since it takes time and effort to set an appropriate concept for pictograms, it is difficult to greatly increase the number of pictograms associated with words. For this reason, at present, it is difficult to sufficiently increase the appearance frequency of pictograms in the connection prediction list.

  The present invention has been made paying attention to the above-mentioned problem, and it is possible to display the connection prediction candidates of pictograms even for the determination of words and phrases that are not associated with pictograms. The challenge is to increase

  The program according to the present invention searches a dictionary in which a plurality of words are registered in association with each connection relationship according to the confirmation of an input target word, and is input subsequent to the confirmed word. Selected in accordance with a prediction means for displaying a list of extracted candidates (hereinafter referred to as “connection prediction list”) and an operation for selecting one candidate in the displayed list. The computer is caused to function as a character input device including candidate determination means for determining and outputting candidates.

  The dictionary of this character input device includes a pictographic prediction dictionary in which a plurality of pictographs are registered in association with conceptual words corresponding to the pictographs. The prediction means includes the pictogram prediction dictionary as a search target, extracts a pictogram corresponding to a phrase that at least partially matches the form of the phrase determined as an input target from the pictogram prediction dictionary, and extracts the extracted pictogram Include in the list above.

  According to the above configuration, when a word / phrase determined as an input target is associated with a pictogram and registered in the pictogram prediction dictionary, a connection prediction list including the pictogram corresponding to the word / phrase can be displayed. . In addition, even if a phrase that is associated with a pictogram and is not registered in the pictogram prediction dictionary is confirmed, if the pictogram is associated with a phrase that matches a part of the form of the phrase, that pictogram is included. A connection prediction list can be displayed.

The predicting means of the present invention has a function of extracting a pictogram associated with a word determined as an input target from a pictogram prediction dictionary and a character string associated with the end or top character string of the word determined as an input target. And a function for extracting a pictograph from the pictograph prediction dictionary. For example, if an emoticon representing the concept of snow is registered in association with the word “snow,” the word that contains “snow” at the end (“snow”) "Every snow" etc.) is confirmed, it is possible to include pictographs representing the concept of snow and include them in the prediction list.

In the character input device according to an embodiment of the present invention, a synonym dictionary in which a plurality of combinations of words having similar concepts are registered is provided, and the predicting unit is similar to a word / phrase determined by searching the synonym dictionary and the pictogram prediction dictionary. A function is added to extract pictograms associated with words having the concept. According to this configuration, it is possible to display a connection prediction list including a pictograph even for a word that does not match the form of the phrase with which the pictograph is associated.

  In the character input device according to another embodiment, the frequency at which a candidate for a pictograph corresponding to a part of the form of a phrase determined as an input target is selected from candidates in the list or the candidate for the pictograph Learning means is included for learning the frequency at which is not selected in association with a combination of a phrase associated with the pictogram in the pictogram prediction dictionary and a confirmed phrase. Further, when the predicting means extracts a pictograph corresponding to a part of the form of the word / phrase determined as an input target by searching the pictograph prediction dictionary, the predicting means associates the word / phrase corresponding to the pictograph with the combination of the confirmed word / phrase. Based on the learned frequency, it is determined whether or not the pictogram is included in the list.

  If the pictograms extracted and connected in the prediction list are partially matched with the confirmed word form and do not conform to the concept of the confirmed word, the pictogram is almost always selected. Absent. According to the above embodiment, since pictograms that have been extracted by the same search in the past but are not selected are excluded from the list, it is possible to prevent pictograms that are not suitable as candidates from being continuously displayed.

The pictograms referred to in the present invention include those composed of font data and those composed of image files. Further, a pictograph according to the present invention includes a combination of a plurality of symbols (such as an emoticon) in order to express some concept.
The pictogram prediction dictionary may be a dedicated dictionary for extracting a pictographic connection prediction candidate, or may be included in a prediction dictionary for extracting a normal character notation connection prediction candidate.

  According to the present invention, even when a phrase that is not associated with a pictogram is confirmed as an input target, it is possible to display a connection prediction list including pictograms of concepts related to the confirmed word. Therefore, character input using pictograms becomes easier and operability is improved, and the interest of character input can be enhanced.

It is a functional block diagram of a character input system to which the present invention is applied, and a diagram showing a data structure of various dictionaries in the system. It is a flowchart which shows the schematic procedure of the process in a character input system. It is a figure explaining the specific example of a synonym matching search. It is a figure explaining the specific example of a tail word matching search. It is a flowchart which shows the procedure of the update process of a non-selection learning dictionary. It is a flowchart which shows the procedure of the creation process of a connection prediction list.

FIG. 1 shows a configuration example of a character input system to which the present invention is applied.
The character input system 1 of this embodiment executes a kana-kanji conversion process in accordance with a user operation in an information processing apparatus such as a mobile phone, and converts the converted character string into an application 100 (notepad, To input to a mailer). Specifically, the system 1 includes a user interface 10, a conversion processing unit 11, a confirmation processing unit 12, a conversion dictionary 13, a connection prediction dictionary 14, a synonym dictionary 15, a learning dictionary 16, a non-selected learning dictionary 17, and the like. All of these are set by incorporating a dedicated program into the information processing apparatus.

  The user interface 10 starts up an operation unit including a character key on a display unit (not shown), and communicates the contents of the key operation to the conversion processing unit 11 and the confirmation processing unit 12, while the process and result (reading character string) Assembly process, candidate list creation process, candidate confirmation, etc.) are displayed. The operation unit is not limited to an image, and an operation unit using real keys may be provided.

  The conversion processing unit 11 cooperates with the user interface 10 to extract candidate words to be input by searching the conversion dictionary 13 and the learning dictionary 16 while assembling a reading character string. The extracted candidates are displayed as a list by the user interface 10. When the user interface 10 accepts an operation for selecting one of the candidates, the confirmation processing unit 12 confirms a character string representing the selected word and outputs it to the application 100.

  The confirmation processing unit 12 is also set with a function for updating the learning dictionary 16 and the non-selected learning dictionary 17 in accordance with candidate confirmation. When the candidate is determined, the conversion processing unit 11 searches the connection prediction dictionary 14 and the learning dictionary 16 by the determined word and extracts candidates for a word (including pictograms) that may be input next. The process (connection prediction process) to perform is performed. By this process, a connection prediction list is created and displayed, and when a candidate in the list is selected, the candidate is confirmed and output to the application 100.

  In FIG. 1, together with the functional block diagram of the system, the schematic data structure of various dictionaries excluding the learning dictionary 16 is shown in a table format ((a) to (d) in FIG. 1). The configuration of each dictionary will be described with reference to these tables.

As shown in FIG. 1A, the dictionary data of the conversion dictionary 13 includes readings of phrases (kana character strings before conversion) and notations (forms after conversion).
Although not shown, the dictionary data of the conversion dictionary 13 includes part of speech type and priority. In addition, dictionary data with pictograms can be registered. In this case, kana notation of a phrase indicating the concept of pictograms is set to be read. Note that pictogram notation is represented by font data or an image file of a single character sentence, but there is also a representation of a combination of a plurality of font data such as a face character.

  In the connection prediction dictionary 14, words that are highly likely to be connected in front and rear are combined and registered. Specifically, as shown in FIG. 1B, dictionary data including the reading and notation of each word is registered for each word combination. In the following, when referring to the front word / phrase, it is referred to as “connection source”, and when referring to the rear word / phrase, it is referred to as “connection destination”.

  In the example of FIG. 1B, dictionary data to which pictograms are connected are extracted and shown. With this type of dictionary data, in principle, the reading of the connected phrase matches the original reading. On the other hand, in the case of dictionary data that is a combination of words in normal character notation, for example, a combination of “always” and “caregiving” or a combination of “good morning” and “you” is generally connected source The reading of and the reading of the connection will be different.

  As shown in FIG. 1C, the synonym dictionary 15 groups a plurality of words and phrases having similar concepts and registers dictionary data for each group. Includes reading and notation.

  The dictionary data stored in the conversion dictionary 13, the connection prediction dictionary 14, and the synonym dictionary 15 are all created in advance by the developer, and are not updated in principle except for some information such as usage frequency. On the other hand, the learning dictionary 16 and the non-selection learning dictionary 17 accumulate dictionary data created in response to selection of candidates by the user, and are frequently updated.

  In the learning dictionary 16, dictionary data based on phrases that are determined according to the user's selection operation and output to the application 100 are accumulated up to a certain number according to the determined order. This dictionary data also includes the reading and notation of the confirmed words.

  The non-selection learning dictionary 17 is used for learning the result of a search for extracting pictographic connection prediction candidates. Specifically, as shown in FIG. 1 (d), for each combination of a fixed phrase, its last word (last word), and an extracted pictogram, these readings and notations and the number of times the pictogram is not selected are calculated. Including dictionary data is created and stored in the non-selected learning dictionary 17. Details of processing related to the dictionary data and use of the dictionary data will be described later.

  FIG. 2 shows a schematic procedure of processing executed by the character input system 1 described above. In the following, the flow of processing associated with various operations will be described with reference to step codes in this figure.

In this process, while accepting an hourly key operation (step S1), a process corresponding to the content of the operation is executed.
First, when an input of a kana indicating reading is accepted ("YES" in step S2), a reading character string before conversion is assembled (step S6), the conversion dictionary 13 is searched using the reading character string, and the reading character is read. A prediction candidate list is created by extracting words having readings that coincide with the column (step S7). Furthermore, it progresses to step S12 and the produced prediction candidate list is displayed.

  When a conversion operation is received in a state where the reading character string is assembled (step S3 is “YES”), the conversion dictionary 13 and the learning dictionary 16 are searched using the reading character string to create a conversion candidate list. (Step S8) After that, the process proceeds to Step S12 to display each candidate in the list. In step S8, a word / phrase partially matching the reading character string is extracted, and a converted character string is created by dividing the reading character string into a head part and a tail part based on the part of speech type of the extracted word / phrase. To do.

When any candidate in the list is selected in a state where the prediction candidate list or the conversion candidate list is displayed (step S4 is “YES”), the selected candidate is fixed and output to the application 100. (Step S9). Further, the learning dictionary 16 and the non-selection learning dictionary 17 are updated in accordance with the confirmation (step S10), and a prediction list (connection prediction list) of words to be input next to the confirmed words is created (step S11). .
Thereafter, in step S12, each candidate in the connection prediction list is displayed. Even when a candidate in this list is selected, step S4 becomes “YES”, and steps S9, S10, S11, and S12 are executed in order. That is, when a candidate in the connection prediction list is selected and the selected word is confirmed, the display of the connection prediction list is updated.

  When a termination operation is performed at a predetermined time (step S5), the character input process is terminated by erasing the display on the operation unit.

  The above-described procedure is almost the same as that in the conventional character input system. However, in step S11 shown by a thick frame, processing of contents different from the conventional one is executed in order to extract pictographic connection prediction candidates. The previous step S10 includes a learning process for increasing the accuracy of pictographic connection prediction candidates. Hereinafter, these processes will be described in detail.

  Pictograph connection prediction candidates (hereinafter referred to as “pictogram candidates”) are extracted by searching the connection prediction dictionary 14, but in conventional search, dictionary data in which the reading and notation of a definite phrase are set as the connection source. Is just extracted. Therefore, unless the pictographs connected to the fixed phrase are registered, the connected pictograph candidates cannot be displayed.

On the other hand, in this embodiment, when the pictograph connected to the fixed phrase is not found, the search shown in FIG. 3 or FIG. 4 is further executed.
3 and 4 show the display development on the character input screen in association with dictionary data related to the display.

In the search process shown in FIG. 3, first, the synonym dictionary 15 is searched by reading and writing a fixed phrase, and a phrase (hereinafter referred to as “synonym”) similar in concept to the fixed phrase is extracted. Then, dictionary data in which the reading and notation of the connection source completely match the reading and notation of the synonym is extracted from the connection prediction dictionary 14. Hereinafter, this search is referred to as “synonymous search”.
When the corresponding dictionary data is found by the synonym match search, pictograms in the data are connected and set as pictograph candidates.

In the specific example shown in FIG. 3, the word “Daisetsu” is confirmed in response to the input of the reading character string “Oyuki”, and a connection prediction list including pictogram candidates is displayed according to the confirmation (in FIG. 3). (A) → (B)), a pictograph candidate is selected ((C) of FIG. 3).
In this example, the connection prediction dictionary 14 does not register a pictograph directly connected to “Daisetsu”, but “snow” is registered in the synonym dictionary 15 as a synonym of “Daisetsu”, and “snow” is stored in the connection prediction dictionary 14. The emoji used as the connection source is registered. Therefore, “snow” is extracted from the synonym dictionary 15 as a synonym of “large snow”, and a pictogram based on “snow” is extracted by executing a synonym matching search using this “snow”. A connection prediction list as shown in 3 (B) is displayed.

In the search processing shown in FIG. 4, from the dictionary data whose connection destination is the pictogram in the connection prediction dictionary 14, the one whose reading and notation of the connection source completely matches the reading and notation of the word at the end of the fixed word phrase is extracted. To do. Hereinafter, a search by this method is referred to as “end word match search”, and a word serving as a search key is referred to as “end word”.
When the corresponding dictionary data is found by the tail word matching search, the pictogram in the data is set as a pictograph candidate.

In the specific example shown in FIG. 4, the word “powder and snow” is confirmed in response to the input of the reading character string “Konayuki”, and a connection prediction list including pictographic candidates is displayed according to the confirmation (FIG. 4). (A) → (b)), a pictograph candidate is selected ((c) of FIG. 3).
The connection prediction dictionary 14 in this example does not register pictographs directly connected to “powder snow”, but includes dictionary data of pictographs connected to “snow” at the end of “powder snow”. Therefore, as a result of extracting the pictographs connected to “snow” as the pictograph candidates by the end word match search using “snow” as a key, a connection prediction list as shown in FIG. 4B is displayed.

  In the example of FIG. 4, since the fixed word is composed of two characters, only the last character “snow” functions as the last word. However, when the fixed word has three or more characters, the last character is changed. A search for the connection prediction dictionary 14 may be performed using a plurality of character strings (character strings from the second character onward as viewed from the beginning to the end) as the end words. However, the character string that becomes the end word needs to function as an independent word by itself.

  According to the search method of FIG. 3 or FIG. 4, even when a word that is not registered as a connection source of pictograms is determined, a pictogram having a concept close to that word is extracted, and a connection prediction list including the pictogram is obtained. Can be created. Thereby, the frequency with which the connection prediction list including the pictogram candidates is displayed according to the determination of the phrase by the normal character notation can be increased, and the convenience can be improved.

  However, according to the end word match search, there is a possibility that pictograms of a concept that cannot be said to be close to a fixed word phrase are extracted. For example, when the phrase “Japan” is confirmed, it is desirable to display pictograms such as Hinomaru and a Japanese map, but such pictograms are combined with “Japan” and registered in the connection prediction dictionary 14. If there is not, the third dictionary data from the top in FIG. 1B is hit by the end word matching search using “book” as the end word of “Japan”, and the pictogram representing the concept of “book” Will be extracted.

  Thus, since there is a slight problem in accuracy in the tail word match search, in this embodiment, this problem is dealt with by two methods. The first method gives priority to the process of searching the dictionary data based on the fixed phrase itself (hereinafter referred to as “exact match search”), and if the exact match search cannot extract pictograms, the synonym search is performed. When the pictograph cannot be extracted even by the synonym search search, the end word match search is executed.

The second method learns the user candidate selection status for the search result of the tail word match search, and even if a pictograph is extracted by the tail word match search, the frequency at which the pictograph is selected in the same past search is low. In this case, the pictograph is not set as a pictograph candidate. The non-selection learning dictionary 17 is a learning dictionary for carrying out this method, and for each combination of a definite word, an end word, and an extracted pictogram when an end word match search is performed, Dictionary data including the number of readings and notations and the number of times that a pictograph candidate is not selected (number of times of non-selection) is created and stored.
When the conversion processing unit 11 performs a tail word matching search, the conversion processing unit 11 reads out dictionary data that matches the content of the search from the non-selection learning dictionary 17, and the number of times of non-selection in the data falls below a predetermined threshold value. In addition, a connection prediction list including pictograms extracted by the search is created. On the other hand, when the number of non-selections is equal to or greater than the threshold, the pictograms extracted by the search are connected and excluded from the prediction list.

  For example, when the above threshold value is set to 5 times and a word match search using “book” at the end is performed in accordance with the determination of the word “Japan”, FIG. According to the data structure of the non-selection learning dictionary 17 shown in d), since the number of non-selection of pictograms extracted by the search has already exceeded the threshold value, the pictograms are not included in the connection prediction list.

  According to the above method, a pictograph with a concept different from the fixed phrase is initially displayed as a candidate for the pictograph, but if there is no opportunity for the user to select it, it will be deceived naturally, and even if the same phrase is fixed, the pictograph candidate It will not be adopted as. Therefore, it is possible to prevent a situation in which a pictograph that does not conform to the concept of the fixed phrase is called as a pictograph candidate.

However, since the learning of the number of times of non-selection of pictograms extracted in the tail word match search is performed for each search content that matches, the pictograms that are excluded from the linked prediction list for a certain fixed phrase However, other definite words can be included in the connection prediction list.
For example, when the word “Japan” is confirmed, even if the “book” pictogram is not displayed as a candidate for the pictogram, the final word match search for the word “gorgeous book” Pictographs can be connected as pictograph candidates and included in the prediction list.

FIG. 5 shows a procedure of processing for updating the non-selected learning dictionary in accordance with the word / phrase confirmation. This process is included in step S10 in FIG.
In the first step S101, it is checked whether or not a pictograph candidate is extracted by end word match search. If this pictograph candidate has not been extracted, step S101 is “NO”, and the process is immediately terminated.

If a pictograph candidate is extracted by the tail word match search ("YES" in step S101), it is checked whether dictionary data corresponding to the search content is registered in the non-selection learning dictionary 17 (step S102). ). If the dictionary data is not registered (“NO” in step S102), the process proceeds to step S103, and dictionary data is created for the combination of the fixed word / phrase, the last word used for the search, and the extracted pictogram. Is stored in the non-selection learning dictionary 17 (step S103). The dictionary data includes readings and notations of each of the definite words, end words, and pictograms, and stores an initial value of 0 as the number of non-selections.
If dictionary data corresponding to the content of the tail word match search has already been registered in the non-selection learning dictionary 17 (step S102 is “YES”), step S103 is skipped.

  Next, in step S104, it is checked whether a pictograph candidate extracted by the end word match search is selected. If a pictograph candidate has not been selected (“NO” at step S104), the process proceeds to step S105, and the number of times of non-selection of dictionary data corresponding to the search contents in the non-selection learning dictionary 17 is increased by one. On the other hand, when a pictograph candidate is selected (step S104 is “YES”), the number of times of non-selection of data corresponding to the search contents in the non-selection learning dictionary 17 is cleared to zero.

FIG. 6 shows a processing procedure (details of step S11 in FIG. 2) regarding the creation of the connection prediction list.
In the first step S201, the connection prediction dictionary 14 and the learning dictionary 16 are subjected to a search by reading and writing a definite phrase, and a connection prediction list including candidates by normal character notation is created.

  Steps S202 to S212 correspond to processing for extracting connection prediction candidates using pictograms. First, in step S202, an exact match search using a fixed word is executed, and when a pictograph is found by this search (step S203), the extracted pictograph is added to the linked prediction list (step S212).

  If no pictogram is found in the exact match search (step S203 is “NO”), the synonym dictionary 15 is searched to extract a synonym for the confirmed phrase (step S204). If a synonym is found by this search (“YES” in step S205), a synonym match search using the synonym is executed (step S206). If a pictograph is found by this search (“YES” in step S207), the process proceeds to step S212, and the extracted pictograph is added to the connection prediction list.

  If no synonym for the fixed phrase is found (“NO” in step S205), or if a synonym is found but no pictograph connected to the synonym is found (“NO” in step S207), the final word of the fixed phrase Is used to execute the end word match search (step S208). When pictograms are extracted by this search (“YES” in step S209), the non-selection learning dictionary 17 is searched by a combination of the fixed phrase, the last word used for the search, and the extracted pictograms. The corresponding number of non-selections is acquired (step S210). If the number of times of non-selection is less than the threshold (step S211 is “YES”), the process proceeds to step S212, and the extracted pictogram is connected and added to the prediction list.

As described above, when pictographs are extracted by any one of the complete match search, the synonym match search, and the end word match search, a process of connecting the pictographs to the prediction list (step S212) is executed, and then the process proceeds to step S213. In addition, when no pictograph is extracted in any of the searches (“NO” in step S209), or the pictograph is extracted by the last word search, the number of non-selection acquired in step S210 has reached the threshold value. In the case (“NO” in step S211), the process proceeds to step S213 without executing step S212.
In step S213, for all candidates including candidates in character notation, processing such as integrating candidates overlapping in the connection prediction list and organizing priorities is executed. The connection prediction list creation process ends with the end of step S213, and the process proceeds to a process of displaying the created list (step S12 in FIG. 2).

  In the above embodiment, priorities are set in the order of exact match search, synonym match search, and end word match search, and when a pictograph is extracted in a search with high priority, no other search is performed. However, the algorithm is not limited to this. For example, all three types of searches may be executed. However, it is desirable to display the pictograms extracted by each search so that the ones extracted by the high priority search are higher.

  Regarding the handling of pictograms extracted by tail word matching search, in the above embodiment, when the number of non-selection exceeds the threshold value, it is excluded from the prediction candidate list. Counting and pictograms for which the number of selections is equal to or less than a threshold value may be excluded from the prediction candidate list.

  However, even pictograms with a low selection frequency may be left in the list by lowering the rank without being excluded from the connection prediction list. For example, when the number of non-selections is less than or equal to a predetermined value, the rank may be set higher in the connection prediction list, and after the number of non-selections exceeds a predetermined value, the rank may be lowered as the number increases.

  Further, instead of the end word match search or together with the end word match search, a search for extracting a pictograph corresponding to the head portion of the confirmed phrase may be executed. By this search, for example, a pictogram associated with “Sakura” is displayed when the word “Sakura Fubuki” is confirmed, or “Book” is displayed when the word “Bookstore” is confirmed. It is possible to display the associated pictogram.

  In the above embodiment, only the pictograms extracted by the tail word match search are determined based on the past learning history as to whether or not to adopt them as pictogram candidates. Similar learning and determination may be performed on the extracted pictograms. In this way, when the user has no habit of using pictograms, it is possible to reduce the frequency at which pictogram candidates are included in the connection prediction list, and to easily display candidates in normal text notation.

The character input system 1 of the above embodiment is set in the information processing apparatus by installing a dedicated program. However, if there is an existing character input system in the apparatus and a plurality of dictionary data to which pictograms are connected are registered in the connection prediction dictionary, an update for performing the processing of FIG. 5 and FIG. You may install the program.
Further, the connection prediction dictionary 14 may be stored separately into a dictionary file having a normal character notation as a connection destination and a dictionary file having a pictogram as a connection destination.

  Finally, the character input system 1 is not limited to a portable information processing apparatus, but can be incorporated in a stationary apparatus (such as a display apparatus or a facsimile). It can also be applied to a personal computer. Further, the input target language is not limited to Japanese, and input of other languages such as English can be handled by a system having the same configuration.

DESCRIPTION OF SYMBOLS 1 Character input system 10 User interface 11 Conversion process part 12 Confirmation process part 13 Conversion dictionary 14 Connection prediction dictionary 15 Synonym dictionary 16 Learning dictionary 17 Non-selection learning dictionary 100 Application

Claims (4)

A dictionary in which a plurality of words are registered in association with each connection relationship is searched according to the confirmation of the input target word, and candidate words to be input subsequent to the determined word are obtained. A character input device comprising: a prediction unit that extracts and displays a list of extracted candidates; and a candidate determination unit that determines and outputs a candidate selected according to an operation of selecting one candidate in the displayed list As a program for causing a computer to function as
The prediction means includes a pictographic prediction dictionary in which a plurality of pictographs are registered in association with concept words corresponding to the pictographs, and includes the pictographs extracted by the search in the list. And a function for extracting a pictograph associated with a word determined as an input target from the pictographic prediction dictionary and a character string at the end or top of the word determined as the input target A program for character input comprising a function of extracting a pictograph from the pictograph prediction dictionary . The predicting means extracts pictograms associated with a phrase having a concept similar to the confirmed phrase by searching for a synonym dictionary in which a plurality of combinations of phrases of similar concepts are registered and the pictogram prediction dictionary The character input program according to claim 1 , further comprising a function of The candidate confirming means is a frequency at which a pictogram candidate corresponding to a part of a form of a phrase that has been confirmed as an input target is selected from the candidates in the list, or a frequency at which the pictogram candidate is not selected. Learning means for learning in association with a combination of a phrase that is associated with the pictogram in the pictogram prediction dictionary and the determined phrase;
When the predicting means extracts a pictograph corresponding to a part of the form of the phrase determined as an input target by searching the pictographic prediction dictionary, it corresponds to a combination of the phrase corresponding to the pictograph and the determined phrase. association based on the frequency that is learned, it is determined whether inclusion of the pictogram in the list, according to claim 1 or character program for input as described in 2. A dictionary in which a plurality of words are registered in association with each connection relationship, and a word that is input following the confirmed word by searching the dictionary in response to confirmation of the word to be input A predictor that extracts a candidate and displays a list of extracted candidates, and a candidate determiner that determines and outputs a candidate selected according to an operation of selecting one candidate in the displayed list. An information processing apparatus,
The dictionary includes a pictogram prediction dictionary in which a plurality of pictograms are registered in association with concept words corresponding to the pictograms, respectively.
The predicting means includes a pictogram extracted by searching the pictogram prediction dictionary in the list, and extracts a pictogram associated with a word / phrase determined as an input target from the pictogram prediction dictionary; An information processing apparatus comprising: a function of extracting a pictograph associated with a character string at the end or top of a word determined as the input target from the pictographic prediction dictionary .

Images