JPH04295962A - Morpheme analyzing device - Google Patents

Abstract

PURPOSE:To offer the morpheme analyzing device for generating and analyzing a new morpheme in which a part of speech of the morpheme is varied by using a part-of-speech variation table. CONSTITUTION:In the morpheme analyzing device which is provided with an analysis control part 1, a morpheme dictionary 2, a part-of-speech variation table 3, and a connection table 4, converts a KANJI (Chinese character) and KANA (Japanese syllabary) mixed sentence to a morpheme train, generates a new morpheme in which a part of speech of the morpheme is varied in accordance with its description by using the part-of-speech variation table 3, and continues an analysis, a subsequent part-of-speech varying means in which a part-of-speech variation related to the subsequent part of speech in the case two parts of speech are adjacent to each other is provided in the part-of-speech variation table.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a morphological analysis device relating to a grammatical connection dictionary in morphological analysis.

0002

[Prior Art] As a conventional morphological analysis device,
For example, as disclosed in Japanese Unexamined Patent Publication No. 1-269168, there is a method that only processes verb conjunctions with a limited scope, and as disclosed in Japanese Unexamined Patent Publication No. 1-236360, There is a method that only processes adjective prefixes. That is, more specifically, the former is
The present invention relates to a sentence analysis device for Japanese sentences that are not separated, and in particular detects words when a verb or adjective changes to a noun, and grammatically accurately analyzes the target sentence. The latter relates to Japanese sentence processing methods, especially morphological analysis of collocations with adjective prefixes.

0003

[Problem to be Solved by the Invention] Among the Japanese morphological analysis methods described above, a morpheme dictionary is looked up based on a Japanese character string, and the morphemes that can be looked up in the dictionary are
There is a method of obtaining a morpheme sequence that is a solution by using a "connection table" that describes the possibility of grammatical connections between parts of speech of the morphemes.

[0004] However, this method can only use binary relations for morphological analysis, and special grammatical matters, such as "
Even though the morpheme "registration" does not have an adjective-like attribute by itself, it acquires an adjective-like attribute by adding the prefix "mi". Therefore, it is difficult to handle cases where the same "registration" has different attributes of connection to the subsequent morpheme depending on whether "not" is added before the word.

[0005]

[Means for Solving the Problems] The invention according to claim 1 includes an analysis control section, a morpheme dictionary, a part-of-speech change table,
a morphological analysis device that converts a sentence containing kanji and kana into a morpheme string, generates a new morpheme by changing the part of speech of the morpheme according to the description using the part of speech change table, and continues analysis. A post-part-of-speech change means is provided in the change table, in which a part-of-speech change regarding a subsequent part-of-speech when two parts of speech are adjacent to each other is described.

[0006] In the invention according to claim 2, an analysis control section;
It is equipped with a morpheme dictionary, a part-of-speech change table, and a connection table, converts a sentence containing kanji and kana into a morpheme string, and uses the part-of-speech change table to generate and analyze a new morpheme in which the part of speech of the morpheme is changed according to its description. In the morphological analysis device that continues the above, the part-of-speech change table is provided with a previous part-of-speech change means that describes a part-of-speech change regarding the previous part-of-speech when two parts of speech are adjacent.

[0007] In the invention according to claim 3, an analysis control section;
It is equipped with a morpheme dictionary, a part-of-speech change table, and a connection table, converts a sentence containing kanji and kana into a morpheme string, and uses the part-of-speech change table to generate and analyze a new morpheme in which the part of speech of the morpheme is changed according to its description. a part-of-speech merging information determining means in which the part-of-speech change table describes information on whether to merge two adjacent parts of speech when two adjacent parts of speech are to be merged; When merging two parts of speech, a part-of-speech merging means is provided that describes a new part of speech that is generated as a result of merging two parts of speech.

[0008] In the invention according to claim 4, an analysis control section;
It is equipped with a morpheme dictionary, a part-of-speech change table, and a connection table, converts a sentence containing kanji and kana into a morpheme string, and uses the part-of-speech change table to generate and analyze a new morpheme in which the part of speech of the morpheme is changed according to its description. In the morphological analysis device that continues the above, the part-of-speech change table contains information regarding the operation of merging the two adjacent parts of speech, changing the subsequent part of speech, or doing nothing when two parts of speech are adjacent. The part-of-speech merging information determination means described, the part-of-speech merging means that describes the newly generated part of speech as a result of merging two parts of speech when merging parts of speech, and the next part of speech when changing the next part of speech. After the part-of-speech change is described, a part-of-speech change means is provided.

[0009] In the invention according to claim 5, an analysis control section;
It is equipped with a morpheme dictionary, a part-of-speech change table, and a connection table, converts a sentence containing kanji and kana into a morpheme string, and uses the part-of-speech change table to generate and analyze a new morpheme in which the part of speech of the morpheme is changed according to its description. In the morphological analysis device that continues the process, the part-of-speech change table contains information regarding actions such as merging the two adjacent parts of speech, changing the previous part of speech, or doing nothing when two parts of speech are adjacent to each other. The part-of-speech merging information determination means described, the part-of-speech merging means that describes the newly generated part of speech as a result of merging two parts of speech when merging parts of speech, and the previous part of speech when changing the previous part of speech. A pre-part-of-speech change means is provided in which the part-of-speech change is described.

0010

[Operation] In the invention as claimed in claim 1, by preparing a part-of-speech change table that describes the part-of-speech change regarding the next part of speech when two parts of speech are adjacent, the connection of the preceding morpheme to the succeeding morpheme is established. It is possible to deal with grammars in which the attributes of the table change, and since it only deals with binary relationships like the connection table, it can be implemented with simple processing.

In the invention according to claim 2, whereas the invention according to claim 1 described above is suitable for morphological analysis processing proceeding from left to right, here, morphological analysis processing proceeding from right to left is suitable. It is suitable for processing, and its effects are similar to those of the invention described in claim 1.

In the invention as claimed in claim 3, by preparing a part-of-speech change table that describes information regarding the merging of two parts of speech when two parts of speech are adjacent to each other,
It is possible to correspond to a grammar in which the preceding and succeeding morphemes apparently become one and have a new grammatical meaning, and moreover,
Like the connection table, it only deals with binary relationships, so it can be implemented with simple processing.

In the invention as claimed in claim 4, the processing can be somewhat simplified by merging two parts of speech when two parts of speech are adjacent to each other, or by preparing a part of speech change table in which part of speech changes related to subsequent parts of speech are described. Although it is complicated, it is possible to obtain the same effects as in the case of the invention described in claims 1 and 3 described above.

In the invention as claimed in claim 5, the processing is somewhat complicated by merging two parts of speech when two parts of speech are adjacent to each other, or by preparing a part-of-speech change table that describes part-of-speech changes related to the previous part of speech. However, the same effects as in the case of the inventions described in claims 2 and 3 described above can be obtained.

[0015]

[Embodiment] An embodiment of the invention according to claim 1 is shown in FIGS. 1 to 4.
The explanation will be based on. FIG. 1 shows the overall configuration of a morphological analysis device, and this device includes an analysis control section 1, a morphological dictionary 2, a part-of-speech change table 3, and a connection table 4. In this case, in the part-of-speech change table 3, as shown in FIG.
is provided. The analysis control section 1 includes a morpheme dictionary lookup section 1a, a part-of-speech change processing section 1b, and an inter-morpheme link generation section 1c.

[0016] In such a configuration, the morphological analysis device converts a sentence containing kanji and kana into a morpheme sequence, uses the part-of-speech change table 3 to generate a new morpheme in which the part of speech of the morpheme is changed according to the description, and analyzes it. Hereinafter, a specific example of performing morphological analysis using the part-of-speech change table 3 having the post-part-of-speech change means A will be described.

[0017] Now, the Japanese character string "He shows his running."
Assume that the example sentence has been input. First, the morphological dictionary lookup section 1a performs a morphological dictionary lookup of the entire Japanese character string,
Morphological candidates 5 as shown in FIG. 3 are obtained. Next, the part-of-speech change processing unit 1b checks the part-of-speech change table 3 for adjacent morpheme candidates in order from the one located on the left side, and if there is a part-of-speech change, a new morpheme in which the part of speech of the subsequent morpheme is changed according to the description is generated. generate. Here, in part-of-speech change table 3, the verb stem ``run'' is followed by the verb conjunctive ending ``ri''.
Since the nounization affix is described as the part-of-speech change when `` is added, the nounification affix ``ri'' is generated as a morpheme candidate. Finally, the inter-morpheme link generation unit 1c examines the connections of the obtained morpheme candidates to later morpheme candidates in order from the one located on the left using the connection table 4, and links between connectable morpheme candidates. and outputs the morpheme candidate group and the link group as a morpheme lattice that becomes the solution. The morpheme lattice 6 thus obtained is shown in FIG.

As described above, by preparing the part-of-speech change table 3 having the post-part-of-speech change means A in which the part-of-speech change related to the next part of speech is described when two parts of speech are adjacent, it is possible to It becomes possible to support grammars in which the attributes of connections to morphemes change. Further, in this case, like the connection table 4, only binary relationships are handled, so it can be realized with simple processing.

Next, an embodiment of the invention according to claim 2 will be explained based on FIGS. 5 to 7. Note that the description of the same parts as in the embodiment of the invention (see FIG. 1) of claim 1 will be omitted, and the same parts will be denoted by the same reference numerals.

[0020] Here, the part-of-speech variation table 3 includes
As shown in FIG. 2, a pre-part-of-speech change means B is provided, which describes a change in part-of-speech regarding the previous part-of-speech when two parts-of-speech are adjacent, and this means is used to perform morphological analysis. Now, regarding the same example sentence as the invention claimed in claim 1,
When a new part of speech "collocation word-type nounization suffix" is set and a morpheme dictionary lookup is performed, morpheme candidates 7 as shown in FIG. 6 are obtained. The part-of-speech change processing unit 1b checks the part-of-speech change table 3 for adjacent morpheme candidates not from left to right but from the right side, and if there is a part-of-speech change, it changes the part of speech of the previous morpheme according to the description. Generate new morphemes. Here, in the part-of-speech change table 3, a noun is described as a part-of-speech change when the verb stem ``shu'' is added before the collocational nounization suffix ``ri'', so the noun ``
``Run'' is generated as a morpheme candidate. Finally, for the obtained morpheme candidates, the inter-morpheme link generation unit 1c examines connections to later morpheme candidates using the connection table 4, starting from the one located on the left, and searches for connections between connectable morpheme candidates. Links are created, and a group of morpheme candidates and a group of links are output as a morpheme lattice serving as a solution. The morpheme lattice 8 thus obtained is shown in FIG. In this case, since the nounization suffix is not connected to the verb stem, only the noun "shu" remains as a candidate, thereby making it possible to form a single morpheme sequence.

As described above, while the invention according to claim 1 is suitable for morphological analysis processing proceeding from left to right, here, the part-of-speech change table 3 having the pre-part-of-speech change means B is used. By providing this, it becomes possible to apply it to morphological analysis processing that proceeds from right to left.

Next, an embodiment of the invention according to claim 3 will be explained based on FIGS. 8 to 10. Note that the description of the same parts as in the embodiment of the invention (see FIG. 1) of claim 1 will be omitted, and the same parts will be denoted by the same reference numerals.

[0023] Here, the part-of-speech variation table 3 includes
As shown in FIG. 1, there is a part-of-speech merging information determination means C that describes information on whether or not to merge two adjacent parts of speech when two adjacent parts of speech are to be merged. A part-of-speech merging means D is provided in which a part-of-speech newly generated as a result of merging two parts of speech is described, and morphological analysis is performed using these means.

Now, for the same example sentence as the invention described in claim 1, if a new part of speech "nominalized verb conjunction form" is set and a morpheme dictionary lookup is performed, a morpheme candidate 9 as shown in FIG. 9 is obtained. In this case, the part-of-speech change processing unit 1b checks the part-of-speech change table 3 as shown in FIG. 8 for adjacent morpheme candidates in order from the one located on the left, and if there is part-of-speech merging, merges the preceding and succeeding morphemes according to the description. , generates a new morpheme. Here, the part-of-speech change table 3 describes the nominalized verb conjunction form as the part-of-speech change when the verb ending "ri" is added after the verb conjunction stem "running", so the nominalized verb conjunction form "hasashi" is Generate as a morpheme candidate. Finally, the inter-morpheme link generation unit 1c examines the connections of the obtained morpheme candidates to later morpheme candidates in order from the one located on the left using the connection table 4, and links between connectable morpheme candidates. and outputs the morpheme candidate group and the link group as a morpheme lattice that becomes the solution. The morpheme lattice 10 obtained in this way is shown in FIG.
Shown below.

As mentioned above, the part-of-speech merged information determining means C
By preparing a part-of-speech change table 3 having a part-of-speech merging means D that describes information about merging parts of speech when two parts of speech are adjacent to each other, the preceding and succeeding morphemes apparently become one and create a new grammar. This makes it possible to support grammars that have specific meanings. Also, like the connection table 4, only binary relationships are handled here, so it can be realized with simple processing.

Next, an embodiment of the invention according to claim 4 will be explained based on FIG. 11. Note that the description of the same parts as in the embodiment of the invention (see FIG. 1) of claim 1 will be omitted, and the same parts will be denoted by the same reference numerals.

[0027] Here, the part-of-speech change table 3 includes
As shown in 1, the part-of-speech merging information judgment describes information regarding the action of merging the two adjacent parts of speech, changing the subsequent part of speech, or doing nothing when two parts of speech are adjacent to each other. means E, a part-of-speech merging means F that describes a new part of speech that will be generated as a result of merging two parts of speech when merging parts of speech, and a part-of-speech change that describes the next part of speech when changing the next part of speech. A part-of-speech changing means G is provided.

The part-of-speech change processing unit 1b performs part-of-speech merging processing from left to right. In this case, in Figure 11, each item is divided by a dotted line, and what is written on the right side of the dotted line is the action of change, and "No" means that no change will occur. 'hen' means that the part of speech that follows changes.
"Jo" indicates that the parts of speech before and after are combined. Moreover, what is written on the right side of the dotted line is the part of speech given to the new morpheme when there is an action of change. Also,
The part-of-speech change processing unit 1b examines the part-of-speech change table 3 for adjacent morpheme candidates in order from the one located on the left side,
If there is a change action, the previous and following morphemes are merged according to the description, or the part of speech of the following morpheme is changed to generate a new morpheme. Finally, the inter-morpheme link generation unit 1c sequentially performs the following for the obtained morpheme candidates, starting from the one located on the left:
Connections to subsequent morpheme candidates are checked using the connection table 4, links are created between connectable morpheme candidates, and the morpheme candidate group and link group are output as a morpheme lattice serving as a solution.

As mentioned above, the part-of-speech merged information determining means E
, a part-of-speech change table 3 having a part-of-speech merging means F in which information regarding the merging of two parts of speech when two parts of speech are adjacent to each other is described, and a post-part-of-speech change means G in which a part-of-speech change regarding the subsequent part of speech is described. Although the processing becomes somewhat complicated by preparing the above, it is possible to obtain the same effects as the above-mentioned inventions described in claims 1 and 3.

Next, an embodiment of the invention set forth in claim 5 will be explained based on FIG. 12. Note that the description of the same parts as in the embodiment of the invention (see FIG. 1) of claim 1 will be omitted, and the same parts will be denoted by the same reference numerals.

[0031] Here, the part-of-speech change table 3 includes
As shown in 2, the part-of-speech merging information judgment describes information regarding the action of merging the two adjacent parts of speech, changing the previous part of speech, or doing nothing when two parts of speech are adjacent to each other. Means H, part-of-speech merging means I which describes the part of speech newly generated as a result of merging two parts of speech when merging parts of speech, and part-of-speech change regarding the previous part of speech when changing the previous part of speech. A pre-part-of-speech changing means J is provided.

The part-of-speech change processing unit 1b performs part-of-speech merging processing from right to left. In this case, in Figure 12, each item is separated by a dotted line, but what is written on the right side of the dotted line is the action of change, and "No" means that there is no change. ``odd'' means that the previous part of speech changes.
"Jo" indicates that the parts of speech before and after are merged. Moreover, what is written on the right side of the dotted line is the part of speech given to the new morpheme when there is an action of change. In addition, the part-of-speech change processing unit 1b checks the part-of-speech change table 3 for adjacent morpheme candidates in order from the one located on the left side, and if there is a change action, merges the preceding and succeeding morphemes according to the description,
Alternatively, a new morpheme is generated by changing the part of speech of the previous morpheme. Finally, the inter-morpheme link generation unit 1c examines the connections of the obtained morpheme candidates to later morpheme candidates in order from the one located on the left using the connection table 4, and links between connectable morpheme candidates. and outputs the morpheme candidate group and the link group as a morpheme lattice that becomes the solution.

As mentioned above, the part-of-speech merged information determining means H
, a part-of-speech change table 3 having a part-of-speech merging means I in which information regarding the merging of two parts of speech when two parts of speech are adjacent to each other is described, and a previous part-of-speech change means J in which a part-of-speech change related to the previous part of speech is described. Although the processing becomes somewhat complicated by preparing the above, it is possible to obtain the same effects as in the case of the invention described in claims 2 and 3 described above.

[0034]

[Effects of the Invention] The invention according to claim 1 includes an analysis control section;
It is equipped with a morpheme dictionary, a part-of-speech change table, and a connection table, converts a sentence containing kanji and kana into a morpheme string, and uses the part-of-speech change table to generate and analyze a new morpheme in which the part of speech of the morpheme is changed according to its description. In the morphological analysis device that continues the above, the part-of-speech change table is provided with a post-part-of-speech change means that describes the part-of-speech change for the next part of speech when two parts of speech are next to each other. By preparing a part-of-speech change table that describes part-of-speech changes related to the following part-of-speech, it becomes possible to support grammars in which the attributes of the connection to the following morpheme change depending on the preceding morpheme. Since it only deals with binary relationships, it can be realized with simple processing.

The invention according to claim 2 further comprises an analysis control section, a morpheme dictionary, a part-of-speech change table, and a connection table, converts a sentence containing kanji and kana into a morpheme sequence, and uses the part-of-speech change table to convert the sentence into a morpheme sequence. In a morphological analysis device that generates a new morpheme by changing the part of speech of the morpheme according to the description and continues the analysis, the part of speech change table describes the previous part of speech change regarding the previous part of speech when two parts of speech are adjacent to each other. Since the changing means is provided, the invention according to claim 1 described above is suitable for morphological analysis processing that proceeds from left to right, but here, it is suitable for morphological analysis processing that proceeds from right to left. , it is possible to obtain the same effect as in the case of the invention according to claim 1.

The invention according to claim 3 is provided with an analysis control unit, a morpheme dictionary, a part-of-speech change table, and a connection table, converts a sentence containing kanji and kana into a morpheme sequence, and uses the part-of-speech change table to convert the sentence into a morpheme sequence. In a morphological analysis device that generates a new morpheme by changing the part of speech of the morpheme according to the description and continues to analyze it, if two parts of speech are adjacent to each other in the part of speech change table, whether to merge the two adjacent parts of speech or not. We provided a part-of-speech merging information determination means that describes information on whether or not the part-of-speech is merging, and a part-of-speech merging means that describes the part of speech that is newly generated as a result of merging two parts of speech when merging parts of speech. By preparing a part-of-speech change table that describes the merging of two parts of speech when two parts of speech are next to each other, we can deal with grammars in which the preceding and succeeding morphemes apparently become one and have a new grammatical meaning. Moreover, since it deals with only binary relationships like the connection table, it can be realized with simple processing.

The invention according to claim 4 further comprises an analysis control unit, a morpheme dictionary, a part-of-speech change table, and a connection table, converts a sentence containing kanji and kana into a morpheme string, and uses the part-of-speech change table to convert the sentence into a morpheme sequence. In a morphological analysis device that generates a new morpheme by changing the part of speech of the morpheme according to the description and continues to analyze it, if two parts of speech are adjacent to each other in the part of speech change table, whether to merge the two adjacent parts of speech or not. A part-of-speech merging information determination means that describes information regarding the action of changing the subsequent part of speech or doing nothing, and a part-of-speech that describes a new part of speech that is generated as a result of merging two parts of speech when merging parts of speech. Since we provided a merging means and a post-part-of-speech change means in which the part-of-speech change related to the later part-of-speech is described when changing the later part-of-speech, it is possible to merge two parts-of-speech when two parts-of-speech are adjacent in this way. Alternatively, the process becomes somewhat more complicated by preparing a part-of-speech change table that describes part-of-speech changes related to the later parts of speech.
It is possible to obtain the same effects as in the case of the invention according to claims 1 and 3 described above.

The invention as set forth in claim 5 is provided with an analysis control unit, a morpheme dictionary, a part-of-speech change table, and a connection table, converts a sentence containing kanji and kana into a morpheme sequence, and uses the part-of-speech change table to convert the sentence into a morpheme sequence. In a morphological analysis device that generates a new morpheme by changing the part of speech of the morpheme according to the description and continues to analyze it, if two parts of speech are adjacent to each other in the part of speech change table, whether to merge the two adjacent parts of speech or not. A part-of-speech merging information determination means that describes information regarding the action of changing the previous part of speech or doing nothing, and a part-of-speech that describes a new part of speech that will be generated as a result of merging two parts of speech when merging parts of speech. Since we provided a merging means and a previous part-of-speech changing means in which the part-of-speech change related to the previous part-of-speech is described when changing the previous part-of-speech, it is possible to merge two parts of speech when two parts of speech are adjacent to each other in this way. Alternatively, the process becomes somewhat complicated by preparing a part-of-speech change table that describes part-of-speech changes related to the previous part-of-speech.
The same effects as in the case of the inventions described in claims 2 and 3 described above can be obtained.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the overall configuration of an embodiment of the invention as claimed in claim 1.

FIG. 2 is an explanatory diagram showing a part of speech change table.

FIG. 3 is an explanatory diagram showing morpheme candidates obtained by looking up the entire Japanese character string in a morpheme dictionary.

FIG. 4 is a schematic diagram showing a case where a group of morpheme candidates and a group of links are output as a morpheme lattice serving as a solution.

FIG. 5 is an explanatory diagram showing a state of a part-of-speech change table according to an embodiment of the invention as claimed in claim 2;

FIG. 6 is an explanatory diagram showing morpheme candidates obtained by looking up the entire Japanese character string in a morpheme dictionary.

FIG. 7 is a schematic diagram showing a case where a group of morpheme candidates and a group of links are output as a morpheme lattice serving as a solution.

FIG. 8 is an explanatory diagram showing a state of a part-of-speech change table according to an embodiment of the invention recited in claim 3;

FIG. 9 is an explanatory diagram showing the state of morpheme candidates obtained by performing a morpheme dictionary lookup of the entire Japanese character string.

FIG. 10 is a schematic diagram showing a case where a group of morpheme candidates and a group of links are output as a morpheme lattice serving as a solution.

FIG. 11 is an explanatory diagram showing a state of a part-of-speech change table according to an embodiment of the invention as set forth in claim 4;

FIG. 12 is an explanatory diagram showing a state of a part-of-speech change table according to an embodiment of the invention set forth in claim 5;

[Explanation of symbols]

1 Analysis control section 2 Morphological dictionary 3 Part of speech change table 4 Connection table A Post-part-of-speech change means B Pre-part-of-speech change means C Part of speech merged information determination means D Part of speech merging means E part of speech merged information determination means F Part of speech merging means G Post part of speech change means H Part of speech merged information determination means I Part of speech merging means J Pre-part-of-speech change means

Claims (5)

[Claims] 1. An analysis control unit, a morpheme dictionary, a part-of-speech change table, and a connection table, which converts a sentence containing kanji and kana into a morpheme string, and converts the part-of-speech of the morpheme according to the description using the part-of-speech change table. In the morphological analysis device that generates a new changed morpheme and continues the analysis, a post-part-of-speech change means is provided in which the part-of-speech change table describes a part-of-speech change regarding a subsequent part of speech when two parts of speech are adjacent to each other. A morphological analysis device featuring: 2. An analysis control unit, comprising an analysis control unit, a morpheme dictionary, a part-of-speech change table, and a connection table, converts a sentence containing kanji and kana into a morpheme sequence, and uses the part-of-speech change table to change the part of speech of the morpheme according to its description. In the morphological analysis device that generates a new changed morpheme and continues the analysis, the part-of-speech change table is provided with a previous part-of-speech change means that describes a part-of-speech change regarding the previous part of speech when two parts of speech are adjacent to each other. A morphological analysis device featuring: 3. An analysis control unit, a morpheme dictionary, a part-of-speech change table, and a connection table, which converts a sentence containing kanji and kana into a morpheme sequence, and uses the part-of-speech change table to change the part of speech of the morpheme according to its description. In a morphological analysis device that generates a new changed morpheme and continues the analysis, the part-of-speech change table contains information as to whether or not to merge two adjacent parts of speech when two parts of speech are adjacent to each other. A morphological analysis device comprising: a part-of-speech merging information determining means; and a part-of-speech merging means in which a part of speech newly generated as a result of merging two parts of speech is described when merging parts of speech. 4. An analysis control unit, a morpheme dictionary, a part-of-speech change table, and a connection table, which converts a sentence containing kanji and kana into a morpheme sequence, and uses the part-of-speech change table to change the part of speech of the morpheme according to its description. In a morphological analysis device that generates a new changed morpheme and continues the analysis, if two parts of speech are adjacent to each other in the part of speech change table, the two adjacent parts of speech are merged or the subsequent part of speech is changed. A part-of-speech merging information determination means in which information regarding actions such as ``or nothing'' is described, and 2.
It is characterized by providing a part-of-speech merging means in which a part of speech newly generated as a result of merging two parts of speech is described, and a post-part-of-speech changing means in which a part-of-speech change related to the later part of speech is described when changing a later part of speech. Morphological analysis device. 5. An analysis control unit, a morpheme dictionary, a part-of-speech change table, and a connection table, which converts a sentence containing kanji and kana into a morpheme sequence, and uses the part-of-speech change table to change the part of speech of the morpheme according to its description. In a morphological analysis device that generates a new changed morpheme and continues the analysis, if two parts of speech are adjacent to each other in the part of speech change table, the two adjacent parts of speech are merged or the previous part of speech is changed. A part-of-speech merging information determination means in which information regarding actions such as ``or nothing'' is described, and 2.
A part-of-speech merging means describes a new part of speech that will be generated as a result of merging two parts of speech, and a previous part-of-speech changing means describes a part-of-speech change related to the previous part of speech when changing the previous part of speech. Morphological analysis device.