JP3360617B2 - Numerical information extracting device, numerical information searching device, storage medium storing numerical information extracting program, and storage medium storing numerical information searching program - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a numerical information extracting apparatus and a numerical information retrieving apparatus, and more particularly to a numerical information extracting apparatus and a numerical information retrieving apparatus capable of targeting a document containing an ambiguous numerical expression, and a method for extracting these numerical information. The present invention relates to a device and a storage medium storing a program for realizing a numerical information search device by a computer.

[0002]

2. Description of the Related Art An example of a conventional document retrieval apparatus capable of handling numerical information contained in a document is disclosed in Japanese Patent Laid-Open No. 8-329.
No. 165. This conventional document search device operates as shown in FIG.

That is, when a document to be processed is input (4801 to 4803), a character string having a specific pattern is extracted from the document as numerical data (4804).
A character string based on a certain rule existing before and after the numerical data is extracted as numeric character string data (4805). Noun data is extracted from the numeric character string data, classified for each predetermined item, and associated with the numerical data (4806). The numerical data of each item obtained in this way is tabulated for each predetermined period such as one month (4807), and the tabulated result data is displayed (4803).

To extract numerical data, the text is examined character by character from the beginning,

[0005]

[Outside 1]

[0006] Find combinations of numbers and numbers, or combinations of numbers and "circles". Numbers extracted as strings are converted to numbers.

Another example of a conventional document retrieval apparatus is disclosed in
No. -215041. FIG. 49 shows the configuration of this conventional document search apparatus. Referring to FIG. 49, this document search device includes a search target document holding unit 4901,
Search condition storage unit 4902, search condition reconstruction unit 4903, full-text search condition storage unit 4904, post-processing condition storage unit 4905, full-text search processing unit 4906, full-text search result storage unit 4907, post-processing condition determination unit 49
08, and a search result holding unit 4909.

In this conventional document search apparatus, search conditions are divided into a character string search and other searches including a numerical search. The character string search performs a full-text search, and the search result is compared with the other search results. To narrow down. In the search by numerical values, for example, items and numerical values are extracted from a document as shown in FIG. 50, converted into numerical values, and compared in magnitude.

[0009]

The first problem with these prior arts is that the numerical representation that appears in a document may not point to a single value. In addition to specifying a range like "from A to B", you can also specify only one of the upper or lower limit like "more than" or "at most"
In some cases, approximate numbers are indicated by “about” or “about”. In this case, even if only the numerical string portion is extracted and converted into a numerical value, the information in the original document is lost, which leads to omission of retrieval.

[0010] The second problem is that when a search is performed in which the type of numerical value (eg, amount, length, weight, etc.) is specified, a search error may occur in a specific unit.
The reason is that units representing different types of quantities may have the same name. For example, "pound"
Is used in both weight and currency units. Therefore, when an attempt is made to perform a search by designating a condition relating to an amount of money, there is a possibility that a document including a numerical value of “pound” indicating the weight is searched.

A third problem is that when the type of document and the description format of a numerical value to be searched are limited, the interpretation of the numerical value may not be uniquely determined. The reason is that the original Japanese notation allows ambiguity or the description is omitted. For example, the description "100, 102 yen" can be interpreted as "100,000 yen" or "100 yen and 102 yen". Further, the description "10 to 150,000" can be interpreted as an abbreviation of "100,000 to 150,000" in addition to the interpretation of "10 to 150,000".

A fourth problem is that if numerical information containing ambiguity is to be searched for information corresponding to all possibilities, a large number of incorrect or unnecessary search results are included in the search user. There is a possibility.

An object of the present invention is to provide a numerical information extracting apparatus capable of extracting numerical information from information relating to numerical values contained in a document in view of the above-mentioned problems of the prior art;
An object of the present invention is to provide a numerical information search device capable of specifying a numerical condition and searching a document.

[0014]

SUMMARY OF THE INVENTION A first numerical information extracting apparatus according to the present invention detects an ambiguity of a delimiter of a numeric character string, and extracts a plurality of numeric character strings from the ambiguous numeric character string. Means are operative to select an appropriate break or to select a plurality of breaks.

A second numerical information extracting apparatus according to the present invention comprises: a unit table in which a unit name and a unit type are registered; a type condition table in which conditions for determining the unit type are registered; A unit selection means for determining the type of unit by referring to the type condition table and the text is provided. When a plurality of types are registered in one unit, an operation is performed to select an appropriate type.

A third numerical information extracting apparatus according to the present invention includes a unit table in which a unit name and a hierarchical relationship between units are registered, and a hierarchical determining means for determining the hierarchical relationship between the units. Operate to register a numerical value expressed using units as single numerical information.

A fourth numerical information extracting apparatus according to the present invention includes a selecting means for selecting numerical information according to a type of a unit included in the numerical information, and operates so as to register only necessary numerical information.

A fifth numerical information extracting apparatus according to the present invention includes a range expression table in which a range expression and a type of the range expression are registered;
A range extracting means for determining the type of range expression by referring to the range expression table and text; and operating to create numerical information to which the type of range expression is added.

The sixth numerical information extracting apparatus of the present invention includes a selecting means for selecting numerical information according to the type of range expression included in the numerical information, and operates to register only necessary numerical information.

The seventh numerical information extracting apparatus according to the present invention includes a part of a character string representing a numerical value or an omission interpolating means for supplementing the omission of a unit, and extracts appropriate numerical information from an abbreviated numerical expression. To work.

The first numerical information retrieval apparatus of the present invention includes numerical value registering means for registering a plurality of numerical information when there is an ambiguity in the extraction of a numerical value. No search works.

The second numerical information retrieval apparatus of the present invention includes numerical value registering means for registering a plurality of numerical information when there is an ambiguity in cutting out a numerical value. No search works.

The third numerical information retrieval apparatus of the present invention includes numerical value registration means for registering as a plurality of numerical information when the type of unit is ambiguous. No search works.

The fourth numerical information retrieval apparatus according to the present invention includes a numerical value registering means for registering a numerical value or a unit in a case where the numerical value or the unit is omitted, so that a correct search can be performed even if the omitted expression is present. Operate.

The fifth numerical information search apparatus of the present invention includes a reference table in which necessary numerical information conditions are registered as search results, and operates so as to output only necessary search results.

The sixth numerical information search device of the present invention includes a reference table in which a display method of a search result is registered, and operates so as to output the search result by changing the display method according to the type or ambiguity of the numerical expression. I do.

The seventh numerical information search apparatus of the present invention includes a reference table in which the display order of search results is registered, and operates to output search results in descending order of need.

An eighth numerical information retrieval apparatus according to the present invention includes a reference table in which a method of selecting and displaying a search result is registered, and reference correction means for correcting the reference table by a user. It operates so that selection and designation of display method can be performed.

[0029] A ninth numerical information retrieval apparatus according to the present invention comprises a numerical value registration means for classifying numerical expression range expressions and ambiguous expressions,
A search unit that converts a search condition according to the type and performs a search, and operates to accurately search for a document including a range expression or an ambiguous numerical expression.

[0030]

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

[First Embodiment] Referring to FIG.
The first embodiment of the present invention relates to a document input unit 1, a text holding unit 2, a numerical value extracting unit 3, and a numerical information holding unit.
4, unit determination means 5, range determination means 6, and output means 8.

The unit determining means 5 includes a unit extracting means 51 and a unit table 52.

The range determining means 6 includes a range extracting means 61 and a range expression table 62.

Each of these means operates as follows.

The document input unit 1 writes the contents of the document to be searched into the text holding unit 2.

The text holding means 2 is a means for storing the contents of a document, and is realized by a file device or a memory.

The numerical value extracting means 3 extracts a character string representing a numerical value from the contents of the document stored in the text holding means 2,
The data is converted into numerical data and written into the numerical information holding means 4 together with the character position in the text.

The unit determination means 5 checks the attached unit of the numerical information stored in the numerical information holding means 4,
It is registered in the numerical information holding means 4. For each of the numerical information stored in the numerical information holding means 4, the unit extracting means 51 checks the character string adjacent before and after with reference to the text holding means 2, and finds a unit registered in the unit table 52. For example, it is added to the numerical information of the numerical information holding means 4.
The unit table 52 registers a character string representing a unit and a positional relationship between numerical values.

The range judging means 6 judges whether the combination of the numerical information stored in the numerical information holding means 4 represents a range, and corrects the information in the numerical information holding means 4. The range extracting means 61 includes a text holding means for each of the numerical information stored in the numerical information holding means 4.
The character string adjacent to the front and back is checked with reference to 2 and if there is information having the same unit and registered in the range expression table 62, the numerical information of the numerical information holding means 4 indicates the range. To change. The range expression table 62 registers a character string representing a range and a positional relationship between numerical values.

The output means 8 registers the contents of the numerical information holding means 4 in a search database or index.

The numerical value extracting means 3 will be described with reference to FIG. The numerical value extracting means 3 includes a numerical string extracting means 31, a numerical value discriminating means 32, and a numerical value converting means 33.

The numeral string extracting means 31 is provided in the text holding means 2
Character strings constituting a numeric expression are extracted from the text stored in the. The character strings constituting the numeric expression include expressions indicating a decimal point, a digit separator, a plus / minus sign, and an exponent, in addition to Arabic numerals and Chinese numerals.

The numerical value discriminating means 32 determines whether the character string can be interpreted as a plurality of numerical expressions, and if so, creates a divided character string. If the character string contains characters other than commas and numbers, no judgment is made, and the character string is converted into a numerical value by the numerical value conversion means 33. When only a comma and a number are included, each number string is divided by a comma and each number string is converted into a numerical value by the numerical value conversion means 33. If the first division result is within three digits and the subsequent division results are all three digits, the character string before division is also converted into a numerical value by the numerical value conversion means 33.

The numerical value conversion means 33 converts the character string into a numerical value and registers it in the numerical value information holding means 4. If the value could not be converted, do not register.

Next, the overall operation of the present embodiment will be described with reference to the flowcharts of FIGS.

First, the text is extracted from the document as a character string by the document input means 1 and is stored in the text holding means.
2 (step 501 in FIG. 5). In a document, information that does not need to be searched by numerical values, such as a document title and a creator name, may be included in a distinguishable format, but such a portion does not need to be extracted. Further, even when numerical information is separated from the beginning, such as when a document creation date is registered in a fixed format, it is not necessary to retrieve the numerical information because it can be easily searched as a numerical value.

Next, numbers are extracted from the text stored in the text holding means 2 by the numerical value extracting means 3, converted into numerical values, and registered in the numerical value information holding means 4 (step 502). Generally, one document contains a plurality of pieces of numerical information, and each of them is registered in association with the character position in the text holding unit 2. For a numeric character string having an ambiguous delimiter, a character string with a plurality of delimiters is created and registered by the numerical value discriminating means 32.

Further, the unit associated with each numerical information is extracted by the unit determining means 5 (step 503). The character string before and after the position where the numerical information is extracted is referred to by the text holding unit 2, and if the unit registered in the unit table 52 is found including the positional relationship, the unit is included in the corresponding data of the numerical information holding unit 4. be registered.

Next, the range determining means 6 summarizes the numerical information specifying the range from among the numerical information (step
504). The character string before and after the position where the numerical information is extracted is referred to by the text holding unit 2, and if the unit registered in the range expression table 62 is found including the positional relationship, the corresponding data of the numerical information holding unit 4 is collected. Re-register as one range.

Finally, the contents of the numerical information holding means 4 are output by the output means 8 together with information (document name, document number, etc.) for identifying the document.

Next, effects of the present embodiment will be described.

In this embodiment, the character string extracted by the number string extracting means 31 is configured to be divided by the numerical value discriminating means 32 and converted into a numerical value. Can be output without omission.

(Example) Next, the operation of this embodiment will be described using a specific example.

An example will be described in which the text “Products of $ 123 or more and $ 567 or less are to be purchased” is stored in text holding means 2 by document input means 1.

FIG. 3 shows an example of the registered contents of the unit table 52. In FIG. 3, the positions of the numbers are indicated by squares (□). For example, “□ Kg” indicates that the unit “Kg” is located after the number, such as “100Kg”. "Square meters per second" indicates that a number is located between "every second" and "meter", such as "40.5 meters per second".

FIG. 4 shows an example of the registered contents of the range expression table 62. In FIG. 4, the positions of the numbers including the units are indicated by squares (□). For example, “□ or more and □ or less” indicates that the number is located as “100,000 yen or more and 200,000 yen or less”.

FIG. 6 shows a numerical information holding means according to this operation example.
FIG. 4 is a diagram showing the contents of FIG. In FIG. 6, reference numeral 601 denotes a lower limit value of a numerical range, 602 denotes an upper limit value, 603 denotes a unit attached to a numerical value, 604 denotes a start position of a numerical expression, and 605 denotes an end position of a numerical expression.

From the text, "1"
The two numerical values of "23" and "567" are extracted and registered in the numerical value information holding means 4 (A in FIG. 6). Furthermore, unit determination means
The unit “dollar” is found by 5 and registered in the numerical information holding unit 4 (B in FIG. 6).

Next, a range expression “□ or more and □ or less” is found by the range judging means 6, and the two numerical information shown in FIG. 6B are collectively registered as one piece of information representing the range (FIG. 6C).

If the text includes a numeric string containing a comma, the numeric value discriminating means 32 also registers the divided numeric values. For example, the string "$ 12,345" would be "1234
$ 5 "," 12 "and" $ 345 "are registered (Fig. 6
D). Commas are usually inserted every three digits, so "1234,56
"7" and "123,45" register only the division result.

It is easy to correspond to a document having a number of separator digits other than 3 or a document in which a character other than a comma is used as a digit separator by changing the number of separator digits or a symbol. It is also possible to register a numerical value indicating the appropriateness of the division method. For example, different appropriateness is given to the divided result and the entire character string (606 in FIG. 6E).

[Second Embodiment] Next, a second embodiment of the present invention will be described.
An embodiment will be described in detail with reference to the drawings.

Referring to FIG. 7, according to a second embodiment of the present invention, a document input unit 1, a text holding unit 2, a numerical value extracting unit 203, a numerical value holding unit 4, a unit determining unit 20,
5, a range determining means 6, and an output means 8.

The unit determining means 205 includes a unit extracting means 251
It includes a unit table 252, a unit selecting means 253, and a type condition table 254.

Each of these means operates roughly as follows.

The numerical value extracting means 203 extracts a character string representing a numerical value from the document content stored in the text storing means 2, converts the character string into numerical data, and writes it into the numerical information storing means 4 together with the character position in the text.

The unit judging means 205 checks the attached unit and the type of the numerical information stored in the numerical information holding means 4 and registers it in the numerical information holding means 4. The unit extracting unit 251 refers to the text holding unit 2 for each of the numerical information stored in the numerical information holding unit 4 and checks adjacent character strings before and after, and searches for a unit registered in the unit table 252. . If there are different types of units with the same name, the type is determined by the unit selection means 253. The unit name and type are added to the numerical information of the numerical information holding means 4. The unit table 252 registers a character string representing a unit, a type of the unit, and a positional relationship between the unit and a numerical value. The unit selection unit 253 receives the character string representing the unit, the corresponding plurality of unit types, and the appearance position of the unit from the unit extraction unit 251, and refers to the text holding unit 2 and the type condition table 254, Determine the type of unit. The type condition table 254 registers conditions for determining the type of unit for each type of unit.

Next, the operation of the unit determination means 205 of the present embodiment will be described with reference to the flowcharts of FIGS.

The unit judging means 205 adds unit information to each of the numerical information registered in the numerical information holding means 4 by the operation described below.

First, the appearance position of a numerical value is read from the numerical value information holding means 4 (step 801 in FIG. 8).

Next, it is checked with reference to the text holding means 2 whether there is a unit registered in the unit table 252 before and after the appearance position of the numerical value (step 802).

If one unit is found, it is registered in the numerical information storage means 4. If a plurality of types of units are found, an appropriate type is determined (step 803).

To determine the type of unit, a condition corresponding to the type of unit found is selected from the selection condition table 254, and an appropriate type is selected with reference to the text holding means 2 (step 804).

Finally, the unit name and unit type are registered in the numerical information holding means 4 (step 805).

Next, effects of the present embodiment will be described.

In this embodiment, since the unit type is registered in the unit table, it is possible to extract numerical information to be used for a search in which the unit type is specified.

Further, in this embodiment, since the unit extracting means is configured to refer to the unit selecting means for judging the type of the unit, the unit extracting means which is appropriate when a unit having the same name but different type competes. You can select the type.

(Example) Next, the operation of this embodiment will be described using a specific example.

FIG. 9 shows an example of the registered contents of the unit table 252. In FIG. 9, in addition to the example of FIG. 3, the type of unit is registered (901 in FIG. 9). For example, “□ yen” represents the amount of money, and “□ Kg” represents the weight. “□ pound” is used to represent both amount and weight.

FIG. 10 shows an example of the registered contents of the type condition table 254. In the example of FIG. 10, the type of unit (10 in FIG. 10)
01) is registered (100 in FIG. 10).
2).

An example is described in which the text “Purchase a product at a price between 123 and 567 pounds” is stored in the text holding means 2 by the document input means 1. FIG. 11 is a diagram showing the contents of the numerical information holding means 4 according to this operation example. In FIG. 11, reference numeral 1106 denotes a unit type.

The numerical value extracting means 3 converts “1” from the text.
Two numerical values of "23" and "567" are extracted and registered in the numerical value information holding means 4 (A in FIG. 11). Further, the unit “pound” is changed to “amount” by the unit extracting means 251.
And “weight” are found and passed to the unit selection means 253 to select the appropriate one. The unit selection means 253 extracts each related word from the type condition table 254,
With reference to the text holding means 2, each word is searched.
In this example, since only “price” registered in the condition of “amount” is found, “amount” is determined as an appropriate type. The unit “pound” and the type “amount” are registered in the numerical information holding means 4 (B in FIG. 11).

When a plurality of related words are found in the unit selecting means 253, they can be selected in the following manner.・ Select the type in which the related word appears closest.・ Select a type that has many related words. -Give importance to the related words, and select the type in which the related word with high importance appears, or the type with a large total of importance.

Another example of the present embodiment will be described.

The unit selecting means 253 obtains the degree of adequacy of each type of unit by a numerical value, and the degree of adequacy is registered in the numerical information holding means 4 together with the type of the unit. An example is shown in FIG. In the numerical value information holding means 4, both types of “money” and “weight” are registered, and the degree of appropriateness is recorded (1107 in FIG. 11). Methods for calculating the degree of appropriateness include the number of related words, the sum of the importance of related words, and a method in which the importance of related words is normalized by the distance from a numeric string.

[Third Embodiment] Next, a third embodiment of the present invention will be described.
An embodiment will be described in detail with reference to the drawings.

Referring to FIG. 12, according to a third embodiment of the present invention, a document input unit 1, a text holding unit 2, a numerical value extracting unit 203, a numerical value storing unit 4, a unit determining unit 30,
5, a range determining means 6, and an output means 8.

The unit determination means 305 comprises: a unit extraction means 351;
It includes a unit table 352 and a hierarchy determination unit 253.

Each of these means operates roughly as follows.

The unit judging means 305 checks the attached unit of the numerical information stored in the numerical information holding means 4 and registers it in the numerical information holding means 4. Unit extraction means 351
For each piece of numerical information stored in the numerical information holding means 4, the text storage means 2 is consulted to check adjacent character strings before and after, and to search for a unit registered in the unit table 252. When a unit is found, the hierarchy determination means 35
According to 3, it is checked whether or not there is a hierarchical structure between adjacent numerical information. If there is a unit having a hierarchical structure, it is combined into one information, and the unit name is added to the numerical information of the numerical information holding means 4. The unit table 352 registers a character string representing a unit, a positional relationship between the unit and a numerical value, and a hierarchical relationship with another unit. The hierarchy determining unit 353 includes a unit extracting unit 351.
And the character string representing the unit and the appearance position of the unit are received, and the unit table 352 and the numerical information holding unit 4 are referred to, and the information in a hierarchical relationship is collected.

Next, the operation of the unit determination means 305 of this embodiment will be described with reference to the flowcharts of FIGS.

The unit determining means 305 adds unit information to each of the numerical information registered in the numerical information holding means 4 by the operation described below.

First, the appearance position of the numerical value is read from the numerical value information holding means 4 (step 1301 in FIG. 13).

Next, it is checked with reference to the text holding means 2 whether there is a unit registered in the unit table 352 before and after the appearance position of the numerical value (step 1302).

For the found unit, the hierarchy determining means 35
The hierarchical relationship with other numerical information is checked by 3 (step 1303).

The hierarchy judging means 353 refers to the numerical information holding means 4, extracts the unit of the numerical information adjacent to the numerical information to be processed at present, and judges the hierarchical relationship by referring to the unit table 352.

If there is no hierarchical relationship, the unit is registered in the numerical information holding means 4 (step 1304).

If there is a hierarchical relationship, numerical information is put together according to the unit relationship (step 1305).

Finally, the unit name and unit type are registered in the numerical information holding means 4 (step 1306).

Next, the effect of the present embodiment will be described.

In the present embodiment, since the hierarchical relationship of units is registered in the unit table, even if one numerical value is expressed using a plurality of units, such as “2 m50 cm”, it is regarded as one numerical value. Can handle.

(Example) Next, the operation of this embodiment will be described using a specific example.

FIG. 14 shows an example of the registered contents of the unit table 352. In FIG. 14, in addition to the example of FIG. 3, unit hierarchical relation information is registered (1401 in FIG. 14). For example, one yen is equal to 100 yen, and 1 m is equal to 100 cm and 0.01 km.

FIG. 15 is a diagram showing the contents of the numerical information holding means 4 according to this operation example.

FIG. 16 is a flowchart showing the operation of the hierarchy determining means 353 of this embodiment.

The text “Deliver a cabinet with a height of 2 m 50 cm” is input by the document input unit 1 to the text holding unit 2.
For example, the case where the information is stored in the memory is described below.

The two numerical values “2” and “50” are extracted from the text by the numerical value extracting means 3 and registered in the numerical value information holding means 4 (A in FIG. 15). Further, the unit "m" is found for the numerical value "2" by the unit extracting means 251. The hierarchy determining means 353 refers to the preceding data in the numerical information holding means 4 (step 1601 in FIG. 16), but since there is no data before, it is determined that there is no hierarchical relationship (step 1602, step 1608).

Next, the unit “cm” is found for the numerical value “50”, and the hierarchy determining means 353 determines the hierarchical relationship.

First, referring to the immediately preceding numerical information, data (B1 in FIG. 15) for the numerical value "2" is obtained (step 1601).

It can be seen from the positional relationship that it is immediately before the numerical value "50", and the hierarchical relationship is determined (step 160).
2).

Referring to the unit table 352 for the unit “cm” corresponding to the numerical value “50”, it is checked whether or not “m” is registered in the hierarchical relationship information. 1603). If it has been registered, the numerical value represented by "m" is converted to "cm" (step 1604).

The unit “m” corresponding to the numerical value “2” is checked with reference to the unit table 352 to see if “cm” is registered in the hierarchical relation information (step 1605).

Since it is registered, "50cm" is replaced by "0.5m"
(Step 1606).

If not, it is determined that there is no hierarchical relationship, and the process ends (step 1608). Converted value "0.5m"
And the previous numerical information “2m” are summed to create numerical information (step 1607).

The created numerical information is registered in the numerical information holding means by the numerical value extracting means 351 (C in FIG. 15).

[Fourth Embodiment] Next, a fourth embodiment of the present invention will be described.
An embodiment will be described in detail with reference to the drawings.

Referring to FIG. 17, according to a fourth embodiment of the present invention, a document input unit 1, a text holding unit 2, a numerical value extracting unit 203, a numerical information holding unit 4, a unit determining unit
205, range determination means 6, information selection means 7, output means 8
It is composed of

The selection means 7 includes a condition determination means 71 and a selection condition table 72.

The condition judging means 71 refers to the selection condition table 72 to extract a part of the contents of the numerical value information holding means 4 and pass it to the output means 8.

Next, effects of the present embodiment will be described.

In this embodiment, since only a part of the contents of the numerical information holding means 4 is selected by the condition judging means 71 and passed to the output means 8, only the necessary kind of numerical information can be selected.

(Example) Next, the operation of this embodiment will be described using a specific example.

FIG. 18 shows an example of the registered contents of the selection condition table 72. In FIG. 18A, the unit of the numerical information passed to the output means 8 is specified. In FIG. 18B, the type of the unit of the numerical information is specified. In FIG. 18C, both the type and the unit of the numerical information are specified. In this example, the amounts are yen, dollar,

All four types and weights are designated.

The condition determining means 71 passes only numerical information having the designated unit to the output means 8.

[Fifth Embodiment] Next, a fifth embodiment of the present invention will be described.
An embodiment will be described in detail with reference to the drawings.

Referring to FIG. 19, according to a fifth embodiment of the present invention, a document input unit 1, a text holding unit 2, a numerical value extracting unit 203, a numerical information holding unit 4, a unit determining unit
5, a range determination unit 506, and an output unit 8.

The range determining means 506 includes a range extracting means 561,
And a range expression table 562.

Each of these means operates roughly as follows.

The range determining means 506 determines whether or not the combination of the numerical information stored in the numerical information holding means 4 represents a range, corrects the information in the numerical information holding means 4, and further converts the range expression. Register the type. Range extraction means
Numeral 561 refers to the text storage unit 2 for each of the numerical information stored in the numerical information storage unit 4 to check adjacent character strings before and after, and has a relationship registered in the range expression table 562. If there is information, the numerical information of the numerical information holding means 4 is changed so as to indicate the range. The range expression table 562 registers a character string representing a range and a positional relationship between numerical values, and also registers a type of range expression.

The operation of range determining means 506 is the same as the operation of range determining means 6 in the first embodiment. However, when registering the numerical information determined to indicate the range in the numerical information holding unit 4, the type of the range expression registered in the range expression table 562 is also registered.

Next, the effect of the present embodiment will be described.

In this embodiment, the range expression table 562
Since the type of the range expression is registered, numerical information suitable for a search device capable of changing a search process depending on the type can be registered. Also, by giving the same type to a range expression having the same meaning, it is possible to register numerical information that does not depend on a difference in expression.

(Example) Next, the operation of this embodiment will be described using a specific example.

FIG. 20 shows an example of the registered contents of the range expression table 562. In FIG. 20, a range expression (2001 in FIG. 20) and a type of range expression (2002 in FIG. 20) are registered. For example, “□ to □” and “□ to □” are both assigned a type of “upper / lower limit”. Different types are assigned to “less than or equal to” and “less than” because they differ in whether or not the numerical value at the position of “□” is included in the range. The type "center" indicates that it is a single numerical value, not a range. The "center" is also given to types of numerical expressions without a range expression.

FIG. 21 shows an example in which numerical information extracted from the text “Products of $ 123 to $ 567 is to be purchased” is registered in the numerical information holding means 4. FIG.
The type of range expression is registered (210 in FIG. 21).
1).

FIG. 22 shows another example of the registered contents of the range expression table 562. In FIG. 22, a range expression and a type of the range expression are registered. The range expression type includes a range taking type (2201 in FIG. 22) and a range ambiguity (220 in FIG. 22).
It is expressed by the combination of 2). In this example, the ambiguity of the meaning of the range is given by a numerical value, 0 if there is no ambiguity,
The larger the ambiguity, the larger the value. A specific value, for example, 0 is given to a numerical expression without an ambiguous expression. For example, “□ or more” and “□□” both indicate the “lower limit” of the numerical value range, but “□□” registers greater ambiguity.

The numerical information extracted from the text “It is expected to be a little over $ 123.”
FIG. 23 shows an example registered in. In FIG. 23, the ambiguity of the range expression is registered (2301 in FIG. 23).

[Sixth Embodiment] Next, a sixth embodiment of the present invention will be described.
An embodiment will be described in detail with reference to the drawings.

Referring to FIG. 24, according to the third embodiment of the present invention, a document input means 1, a text holding means 2, a numerical value extracting means 203, a numerical information holding means 4, a unit determining means 20
5, range determination means 6, information selection means 607, and output means 8
It is composed of

The selection means 607 includes a condition determination means 671 and a selection condition table 672.

The condition judging means 671 includes a selection condition table 672
And extract a part of the contents of the numerical information holding means 4 and pass it to the output means 8.

Next, effects of the present embodiment will be described.

In this embodiment, only a part of the contents of the numerical information holding means 4 is selected by the condition judging means 607,
Since the information is passed to the output unit 8, only necessary information can be selected from the range expression or the ambiguous expression.

(Example) Next, the operation of this embodiment will be described using a specific example.

FIG. 25 shows an example of the registered contents of the selection condition table 672. In FIG. 25, how to select the range of numerical information to be selected (2501 in FIG. 25) and ambiguity (2502 in FIG. 25) are registered. In FIG. 25A, the range is not limited,
You have specified an ambiguity equal to 0. In B of FIG. 25, a method in which the range is set to “center” and the ambiguity is less than 20 is designated.

The condition judging means 671 passes only the numerical information having the designated unit to the output means 8.

[Seventh Embodiment] Next, a seventh embodiment of the present invention will be described.
An embodiment will be described in detail with reference to the drawings.

Referring to FIG. 26, according to the seventh embodiment of the present invention, a document input unit 1, a text holding unit 2, a numerical value extracting unit 203, a numerical information holding unit 4, a unit determining unit 5,
, Range determination means 6, output means 8, and omission interpolation means 9.

The abbreviated interpolation means 9 includes a part of the numerical character string in the document, out of the information registered in the numerical information holding means 4,
Alternatively, the unit not correctly registered because the unit is omitted is corrected with reference to the text holding unit 2.

Next, the operation of the omitted interpolation means 9 of this embodiment will be described with reference to the flowchart of FIG.

The omitting interpolation means 9 performs the following processing on each data registered in the numerical value information holding means 4.

First, it is determined whether or not the number string is omitted (step 2701 in FIG. 27).

If there is a possibility that the number string has been omitted, the omitted number string is estimated by referring to the text holding means 2, and the number string that has not been omitted is restored and digitized (step
2702).

Next, it is determined whether or not the unit is omitted (step 2703).

If there is a possibility of being omitted, the omitted unit is estimated with reference to the text holding means 2 (step 2704).

Finally, the interpolated numerical value and unit are registered in the numerical value information holding means 4 (step 2705).

Next, the effects of the present embodiment will be described.

In the present embodiment, the provision of the omission interpolation means 9 makes it possible to correctly extract numerical information even when a number string or unit is omitted.

(Example) Next, the operation of the omitting interpolation means 9 of this embodiment will be described using a specific example.

FIG. 28 shows an example in which numerical information extracted from the text “Price is expected to be from 10000 to 150,000 yen” is registered in the numerical information holding means 4. A of FIG. 28 shows a state before the processing by the omitting interpolation means 9. The operation of the abbreviated interpolation means 9 for this numerical information will be described.

First, check whether the numerical value is omitted (step 27).
01). With reference to the text holding unit and the numerical information holding unit 4, a list of numerical information that is applicable to the range expression or separated by a comma or a reading point, and only the last numerical information in the list has a unit, and A Chinese numeral string indicating the place ("Hundred", "Thousand", "Million", "Billion", "Trillion", etc., and a combination thereof) attached only to the numeral string in the text corresponding to the last numerical information If there is, supplement the other numbers with the same kanji string. For example, for "1 to 2 billion tons", "10" is supplemented by "100 million". However, “10 to 2 billion tons” does not make up for it. In "two or three million", "two" is supplemented by "one million". The numerical information of A in FIG. 28 applies to the range expression “□ to □” in the text, and “10,000” is added to only the subsequent numerical information, so “10,000” is also added to “10”. And digitized (B in FIG. 28).

Next, it is checked whether or not the unit is omitted (step 2703 in FIG. 27). With reference to the text storage unit and the numerical information storage unit 4, whether the range expression is applicable,
If there is a list of numerical information separated by commas or reading points and only the last numerical information in the list has a unit, the same unit as the last numerical information is given to other numerical information in the list. For example, expressions such as "100 to 200 m" and "25, 26" are applicable. Since the numerical information of B in FIG. 28 applies to the range expression of “□ to □” in the text, the unit “circle” is also registered for “10” (C in FIG. 28).

Finally, the range obtained by supplementing the omission is
Register in the numerical information holding means 4 (step 270 in FIG. 27)
5, FIG. 28D).

Next, the operation of the omitting interpolation means 9 according to another embodiment of the present embodiment will be described.

The omission interpolation means 9 of this embodiment adds the information for distinguishing the omission from other numerical information to the numerical information holding means 4 and registers the result.

FIG. 29A shows an example of the result of additional registration in the numerical information holding means 4. Reference numeral 2901 in FIG. 29 is an area for determining the result of supplementing the omission, and “supplement” is added to the result of supplementing the omission. Further, it is also possible to supplement only the unit without supplementing the numeral string (B in FIG. 29).

[Eighth Embodiment] Next, an eighth embodiment of the present invention will be described.
An embodiment will be described in detail with reference to the drawings.

Referring to FIG. 30, in the eighth embodiment of the present invention, a control means 3001, a numerical value registration means 3002, an information storage means 3003, a condition input means 3004, and a search means 3005
And a result holding means 3006 and a display means 3007.

Each of these means operates roughly as follows.

The control means 3001 controls the entire apparatus.
Specifically, it is realized by a CPU and a memory.

The numerical value registering means 3002 extracts a character string representing a numerical value from the document, converts it into a numerical value, and registers it in the information storage means 3003.

The information storage means 3003 registers numerical information in each document. It is realized as a file device or a database or index on a memory.

The condition input means 3004 inputs a search condition specified by the user of the apparatus.

The search means 3005 searches the information storage means 3003 according to the above search conditions, and registers the result in the result holding means 3006.

The result holding means 3006 is realized by a file device or a memory, and registers search results.

Display means 3007 outputs the contents of result holding means 3006. Output methods include display on a display, printing on a printer, reading aloud by voice, and writing to a file device.

Next, the overall operation of the present embodiment will be described with reference to FIG.

When registering numerical information, control means 3001
Extracts numerical information from the document and registers it in the information storage means 3003 by the numerical value registration means 3002. Numerical expressions in the text may cause missing information or ambiguity in how to extract the numerical expressions due to ambiguity in notation and omission of characters. Also, multiple interpretations may be possible, depending on units having multiple meanings or syntactic ambiguity. The numerical value registration unit 3002 of the present embodiment registers a plurality of interpretation results when a plurality of interpretations occur in the numerical information when extracting and digitizing the numerical information. If there is a difference in the degree of appropriateness among a plurality of interpretations, information indicating the appropriateness is registered at the same time.

When retrieving numerical information, control means 3001
Inputs a search condition from the condition input means 3004. The condition input means 3004 is connected to an input terminal device, and the search condition is input by a mathematical expression, a sentence, a search language, or the like.

Next, the control means 3001 searches the information storage means 3003 by the search means 3005, and stores the result in the result holding means 303.
Register at 06.

Finally, the display means 3007 is changed to the information storage means 3006
And output the result. As a result, only the document number and the document name, part or all of the document, etc. are used.

Next, effects of the present embodiment will be described.

In the present embodiment, when there are a plurality of interpretations of numerical values, since the numerical value registration means 3002 registers them as plural pieces of numerical information, documents containing numerical expressions that cannot be uniquely interpreted are also leaked. You can do a search without any.

Further, with respect to numerical information having a plurality of interpretations, it is possible to output search results which are easy to use, by utilizing information indicating the appropriateness of the interpretation.

[Ninth Embodiment] Next, a ninth embodiment of the present invention will be described.
An embodiment will be described in detail with reference to the drawings.

Referring to FIG. 31, according to the ninth embodiment of the present invention, a control means 3001, a numerical value registering means 3102, an information storing means 3003, a condition inputting means 3004, and a searching means 3005 are provided.
And a result holding means 3006 and a display means 3007.

Referring to FIG. 32, numerical value registration means 3102
Includes a document input unit 1, a text holding unit 2, a numerical value extracting unit 3, a numerical information holding unit 4, a unit determining unit 5, a range determining unit 6, and an information registering unit 3208.

The numerical value registration means 3102 extracts a numerical expression from the text part of the document to be registered, converts it into a numerical value, and registers it in the information storage means 3003. If multiple interpretations can be made for the division of the numeric string, the numeric string is extracted and digitized by each of the multiple divisions.

The information registering means 3208 stores the numerical information holding means 4
Information that determines the contents of the document (such as the document name and document number)
Register with

Next, effects of the present embodiment will be described.

In the present embodiment, when there are a plurality of interpretations of how to divide a numerical value, the numerical value registration means 3102 registers as a plurality of pieces of numerical information. A search without omission can be performed.

(Example) Next, the operation of this embodiment will be described using a specific example.

From the document including the text “$ 12,345”, “$ 12,345”, “12”, and “$ 345” are extracted by the numerical value registering means 3102 (D in FIG. 6).
This numerical information is registered together with the document number (in this example, it is # 1).

From the condition inputting means 3004, if the search condition is "300
When "$ 500 or more and less than $ 500" is input, the retrieval means 3005 extracts numerical information of "$ 345", and the display means 3
007 displays the document name and contents of document # 1. "120
If the condition is $ 0 or more and $ 2,000 or less, "12345
You can search the numerical information of "dollar" and search document # 1. No matter whether the author of the document is "12345" or "12 and 345", no search is missed.

[Tenth Embodiment] Next, a tenth embodiment of the present invention will be described in detail with reference to the drawings.

Referring to FIG. 33, in the tenth embodiment of the present invention, a control means 3001, a numerical value registration means 3302, an information storage means 3003, a condition input means 3004, and a search means 3005
And a result holding means 3006 and a display means 3007.

Referring to FIG. 34, numerical value registration means 3302
Document input means 1, text holding means 2, numerical value extracting means 203, numerical information holding means 4, unit determination means 205,
It includes a range determining means 6 and an information registering means 3208.

The numerical value registering means 3302 extracts a numerical expression from the text portion of the document to be registered, converts it into a numerical value, and registers it in the information storage means 3003 together with the unit and the type of unit. If multiple interpretations are possible for the unit type, numerical information is registered for multiple types.

Next, effects of the present embodiment will be described.

In the present embodiment, when there are a plurality of interpretations of the unit type attached to the numerical value, the numerical value registration means 3302 registers the information as a plurality of numerical information. A search without omission can be performed.

(Example) Next, the operation of this embodiment will be described using a specific example.

According to the numerical value registering means 3302, from the document including the text “Purchase a product at a price of 123 pounds or more and 567 pounds or less”, two numerical values are obtained because “pound” is a unit of money and weight. Is extracted (FIG. 11).
C). This numerical information is registered together with the document number (in this example, # 2).

[0204] Therefore, no matter which of money or weight is designated from the condition input means 3004, there is no leakage from the search object.

[Eleventh Embodiment] Next, an eleventh embodiment of the present invention will be described in detail with reference to the drawings.

Referring to FIG. 35, in the eleventh embodiment of the present invention, a control means 3001, a numerical value registering means 3502, an information storing means 3003, a condition inputting means 3004, and a searching means 3005
And a result holding means 3006 and a display means 3007.

Referring to FIG. 36, numerical value registration means 3502
Is a document input unit 1, a text holding unit 2, a numerical value extracting unit 203, a numerical information holding unit 4, a unit determining unit 5, a range determining unit 6, an abbreviated interpolation unit 9, and an information registering unit 3208. Including.

The numerical value registering means 3502 extracts a numerical expression from the text part of the document to be registered, converts it into a numerical value, and registers it in the information storing means 3003. If there is an abbreviation in a number string or unit,
Numerical information supplementing the omission is created and registered by the omission interpolation means 9.

Next, the effect of this embodiment will be described.

In the present embodiment, for numerical expressions including abbreviations, numerical information supplementing the abbreviations is created and registered, so that documents including the abbreviations can be correctly searched.

(Example) Next, the operation of this embodiment will be described using a specific example.

Numerical information meaning “100,000 to 150,000 yen” is extracted from the document including the text “Price is expected to be from 100,000 to 150,000 yen” by numerical value registering means 3502 (FIG. 28). D). This allows, for example, "1
This information is not searched for under the search condition of "10,000 yen or less".

[0213] In order to avoid omission of search due to omission interpolation error, numerical information before and after the omission can be registered together. At this time, information meaning "10 to 150,000 yen" is also registered from the above text (see FIG. 29).
B). Therefore, no matter which is the intention of the document, it does not leak from the search target.

[Thirteenth Embodiment] Next, a thirteenth embodiment of the present invention will be described in detail with reference to the drawings.

Referring to FIG. 41, in the thirteenth embodiment of the present invention, a control means 3001, a numerical value registration means 3002, an information storage means 3003, a condition input means 3004, and a search means 3005
, Result holding means 3006, display means 3707, and reference correcting means 4108.

The reference correction means 4108 is a means for the user to correct the contents of the reference table 3773. The contents of the current reference table are displayed, and the user is provided with a means for correcting the contents using a text, a check button, or the like.

FIG. 42 is an example of a screen display for correcting the reference table 3773. At the time of correction, there are a method of using a menu for each item (A in FIG. 40) and a method of selecting a combination of some selection conditions (B in FIG. 40).
In addition, there are methods of using a scroll menu or directly using an editor. It is also possible that some of the settings cannot be changed.

Next, effects of the present embodiment will be described.

In the present embodiment, the user can modify the reference table for designating the display target and the display method of the search result by using the reference modifying means, so that the user can select necessary information and an easily viewable form.

[Fourteenth Embodiment] Next, a fourteenth embodiment of the present invention will be described in detail with reference to the drawings.

Referring to FIG. 43, in the thirteenth embodiment of the present invention, a control means 3001, a numerical value registration means 4302, an information storage means 3003, a condition input means 3004, and a search means 4305 are provided.
And a result holding means 3006 and a display means 3007.

Referring to FIG. 44, numerical value registration means 4302
Document input means 1, text holding means 2, numerical value extracting means 203, numerical information holding means 4, unit determination means 5,
It includes a range determination unit 506 and an information registration unit 3208.

Referring to FIG. 45, search means 4305 includes approximate number conversion means 4351, conversion definition means 4352, and information search means
And 4353.

Next, the operation of the present embodiment will be described.

The numerical value registration means 4302 extracts a numerical expression from the text part of the document to be registered, converts it into a numerical value, and registers it in the information storage means 3003. If numbers are represented by ranges,
When an expression representing an ambiguous numerical value is used, the type of the range expression is registered together with the numerical expression.

The condition conversion means 4351 refers to the range definition means 4352 and converts the search condition input from the condition input means 3004 into a condition specification for each type of range.

The range definition means 4352 registers a range of numerical values corresponding to the type of range expression.

The information retrieving means 4353 conducts a search under the conditions converted by the condition converting means 4351, and stores the result in the result holding means 3006.
Register with.

Next, effects of the present embodiment will be described.

In the present embodiment, since approximate number definition means for converting an ambiguous numerical expression into a range of numerical values is provided, it is possible to appropriately search even an ambiguous numerical expression.

(Example) Next, the operation of this embodiment will be described using a specific example.

FIG. 46 is a diagram showing an example of the contents of the approximate number defining means 4352. Reference numeral 4601 in FIG. 46 denotes a type of range expression, which is common to the type given by the range determination unit 506 to numerical information. 460
2 is the range of numerical expression corresponding to the type of range expression, min
And max are the lower limit of the numerical range of the numerical information holding means 4 (see FIG. 6).
601) and the upper limit (602 in FIG. 6).

The operation will be described with an example in which information "30 is included in the range" is retrieved from the numerical information shown in FIG. In FIG.
The document number, the upper limit and lower limit of the numerical value, and the type of range expression are shown, and other information is omitted.

From the condition inputting means 3004, "include 30"
Is entered. The condition conversion means 4351 is converted to the following search formula with reference to the range definition means 4352.

Range type = center and 30 = min or range type = upper limit and 30 ≦ max or range type = upper 2 and 30 <max or range type = ambiguous upper limit and max * 0.8 ≦ 30 and 30 ≦ max or range type = Upper and lower limits and min ≤ 30 and 30 ≤ max As a result of the search using the above search formula, document numbers 2, 3, and 5 are found.

The constant (0.8 in the example of FIG. 46) in the range definition means 4352 can be specified by the user before or during the search. If the type of range expression has a numerical value with ambiguity or adequacy (2202 in FIG. 22), this numerical value may be adjusted (for example, multiplied by a constant or added or subtracted by a certain number) before use. it can.

Although the present invention has been described based on the embodiments, the present invention is not limited to only these embodiments. Further, in order to realize the numerical information extracting device and the numerical information searching device of the above-described embodiment by a computer, the functional means of the present invention is stored in a storage medium typified by a CD-ROM or a floppy disk as a computer-readable program. It may be in a form provided by being stored.

[0238]

The effect of the present invention is not limited to the document format,
The ability to extract and retrieve numerical information from documents containing a wide range of numerical values.

The first effect is that numerical values can be extracted and searched for a range or an ambiguous expression. The reason is that the range expression is classified and processed according to the range and the ambiguity in the range determination means. Another reason is to classify ranges and ambiguous expressions and generate an appropriate search formula according to the type.

The second effect is that various unit expressions can be handled. The reason is that using the hierarchical relationship of units, a number expressed in multiple units can be processed as a single number, and it also supports expressions where some units are omitted. is there.

The third effect is that the amount of data to be output as needed or the amount of search results can be reduced. The reason is that the unit and the range expression can be classified, and only the necessary type of numerical information is selected by the selection means and the display means. The fourth effect is that a correct numerical value can be extracted from a numerical string including ambiguity and omission of a delimiter, and that a search can be performed without fail. The reason is that a numeric value is extracted from a character string containing ambiguous delimiters using a plurality of delimiters. Also, the omitted numeral string is used to supplement the omission from the context.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is a block diagram illustrating a configuration of a numerical value extracting unit 3 according to the first embodiment.

FIG. 3 is a diagram illustrating an example of registered contents of a unit table 52 according to the first embodiment.

FIG. 4 is a diagram illustrating an example of registered contents of a range expression table 62 according to the first embodiment.

FIG. 5 is a flowchart showing the operation of the first embodiment.

FIG. 6 is a diagram illustrating an example of the contents of a numerical information holding unit 4 according to the first embodiment.

FIG. 7 is a block diagram showing a configuration of a second exemplary embodiment of the present invention.

FIG. 8 is a flowchart illustrating an operation of a unit determination unit 205 according to the second embodiment.

FIG. 9 is a diagram illustrating an example of registered contents of a unit table 252 according to the second embodiment.

FIG. 10 is a diagram illustrating an example of registered contents of a type condition table 254 according to the second embodiment.

FIG. 11 is a diagram showing an example of the contents of a numerical information holding unit 4 according to the second embodiment.

FIG. 12 is a block diagram showing a configuration of a third exemplary embodiment of the present invention.

FIG. 13 is a flowchart illustrating an operation of a unit determination unit 305 according to the third embodiment.

FIG. 14 is a diagram illustrating an example of registered contents of a unit table 352 according to the third embodiment.

FIG. 15 is a diagram illustrating an example of contents of a numerical value information holding unit 4 according to the third embodiment.

FIG. 16 is a flowchart illustrating an operation of a hierarchy determining unit 353 according to the third embodiment.

FIG. 17 is a block diagram showing a configuration of a fourth exemplary embodiment of the present invention.

FIG. 18 is a diagram illustrating an example of contents of a selection condition table 72 according to the fourth embodiment.

FIG. 19 is a block diagram showing a configuration of a fifth exemplary embodiment of the present invention.

FIG. 20 is a diagram illustrating an example of registered contents of a range expression table 562 according to the fifth embodiment.

FIG. 21 is a diagram illustrating an example of the contents of a numerical information holding unit 4 according to the fifth embodiment.

FIG. 22 is a diagram illustrating another example of the registered contents of the range expression table 562 according to the fifth embodiment.

FIG. 23 is a diagram showing another example of the contents of the numerical value information holding means 4 according to the fifth embodiment.

FIG. 24 is a block diagram showing a configuration of a sixth exemplary embodiment of the present invention.

FIG. 25 is a diagram illustrating an example of the contents of a selection condition table 672 according to the sixth embodiment.

FIG. 26 is a block diagram showing a configuration of a seventh exemplary embodiment of the present invention.

FIG. 27 is a flowchart showing the operation of the omission interpolation means 9 of the seventh embodiment.

FIG. 28 is a diagram illustrating an example of the contents of a numerical information holding unit 4 according to the seventh embodiment.

FIG. 29 is a diagram showing another example of the contents of the numerical value information holding means 4 of the seventh embodiment.

FIG. 30 is a block diagram showing a configuration of an eighth embodiment of the present invention.

FIG. 31 is a block diagram showing a configuration of a ninth embodiment of the present invention.

FIG. 32 is a block diagram illustrating a configuration of a numerical value registration unit 3102 according to the ninth embodiment.

FIG. 33 is a block diagram showing a configuration of a tenth embodiment of the present invention.

FIG. 34 is a block diagram showing a configuration of a numerical value register 3302 according to the tenth embodiment.

FIG. 35 is a block diagram showing a configuration of an eleventh embodiment of the present invention.

FIG. 36 is a block diagram showing a configuration of a numerical value register 3502 according to the eleventh embodiment.

FIG. 37 is a block diagram showing a configuration of a twelfth embodiment of the present invention.

FIG. 38 is a diagram illustrating an example of the content of a reference table 3773 according to the twelfth embodiment.

FIG. 39 is a diagram illustrating a second example of the content of the reference table 3773 according to the twelfth embodiment.

FIG. 40 is a diagram illustrating a third example of the content of the reference table 3773 according to the twelfth embodiment.

FIG. 41 is a block diagram showing a configuration of a thirteenth embodiment of the present invention.

FIG. 42 is an example of a screen display by the reference correcting unit 4108 according to the thirteenth embodiment.

FIG. 43 is a block diagram showing a configuration of a fourteenth embodiment of the present invention.

FIG. 44 is a block diagram showing a configuration of a numerical value register 4302 according to the fourteenth embodiment.

FIG. 45 is a block diagram showing a configuration of a search unit 4305 according to the fourteenth embodiment.

FIG. 46 is a diagram illustrating an example of the contents of a range definition unit 4352 according to the fourteenth embodiment.

FIG. 47 is a diagram illustrating an example of numerical information according to a fourteenth embodiment.

FIG. 48 is a flowchart showing an operation of an example of a conventional document search device.

FIG. 49 is a block diagram showing the configuration of another example of a conventional document search device.

FIG. 50 is an example of a search target document of another example of the conventional document search device.

[Explanation of symbols]

 1 Document inputting means 2 Text holding means 3, 203 Numerical value extracting means 4 Numerical information holding means 5, 205, 305 Unit judging means 6, 506 Range judging means 7, 607 Information selecting means 8 Output means 9 Omitting interpolation means 3001 Control means 3002 , 3102, 3302, 3502, 4302 Numerical registration means 3003 Information storage means 3004 Condition input means 3005, 4305 Search means 3006 Result holding means 3007, 3707 Display means 4108 Reference correction means

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-7-36917 (JP, A) JP-A-62-284460 (JP, A) JP-A-3-24668 (JP, A) JP-A-6-284 176073 (JP, A) JP-A-3-209564 (JP, A) JP-A-10-27235 (JP, A) JP-A-7-288670 (JP, A) (58) Fields investigated (Int. ⁷ , DB name) G06F 17/30 G06K 9/00

Claims (26)

(57) [Claims] 1. A numerical value information extracting device for extracting a character string representing a numerical value contained in a text, detecting an ambiguous delimiter of a character string representing the numerical value, and extracting a plurality of numerical values from the ambiguous character string. A numerical information extraction device characterized in that a character string is extracted. 2. A numerical value extracting means for extracting a character string representing a numerical value included in a text and generating numerical information obtained by converting the character string into a numerical value, and adding a unit to the numerical information by referring to the text. Unit determination means for performing, a range determination means for adding information representing a range of numerical values to the numerical information or a combination of the numerical information with reference to the text, and an output means for outputting the numerical information. The numerical value extracting means is configured to detect the ambiguity of a delimiter of a character string representing the numerical value and to extract a plurality of numerical character strings from the ambiguous numerical character string. Numerical information extraction device characterized by being performed. 3. A numerical information extracting apparatus for extracting a character string representing a numerical value contained in a text, comprising: a first table in which a unit name and a unit type are registered in advance; and a unit for determining the unit type. The second that registered the condition
When the text and the first table are referred to in order to add a unit to the numerical information to be extracted, a unit having the same name and a different type conflicts. An apparatus for extracting numerical information, wherein an appropriate type is selected by referring to a second table, and a unit is added to the numerical information. 4. A numerical value extracting means for extracting a character string representing a numerical value contained in a text and generating numerical information obtained by converting the character string into a numerical value, and adding a unit to the numerical information by referring to the text. Unit determination means for performing, a range determination means for adding information representing a range of numerical values to the numerical information or a combination of the numerical information with reference to the text, and an output means for outputting the numerical information. The unit determination means includes: a unit table in which unit names and unit types are registered; a type condition table in which conditions for determining the unit types are registered; and a unit type determination unit. If there is a unit having the same name but a different type as a result of referring to the text and the unit table, at least unit selection means for determining the type of the unit by referring to the type condition table is provided. A numerical information extracting apparatus characterized in that a unit is added to the numerical information by being configured as described above. 5. The information processing apparatus according to claim 4, further comprising an information selection unit for selecting the numerical information according to a type of a unit included in the numerical information before outputting the numerical information to the output unit. Numeric information extraction device. 6. A numerical information extracting apparatus for extracting a character string representing a numerical value included in a text, wherein a table in which a range expression and a type of the range expression are registered is prepared in advance, and the text and the table are referred to. A type of range expression, thereby determining the type of range expression, and adding the determined range expression to the extracted numerical information. 7. Numerical value extracting means for extracting a character string representing a numerical value included in a text and generating numerical information obtained by converting the character string into a numerical value, and adding a unit to the numerical information by referring to the text Unit determination means for performing, a range determination means for adding information representing a range of numerical values to the numerical information or a combination of the numerical information with reference to the text, and an output means for outputting the numerical information. The range determining means includes: a range expression table in which a range expression and a type of the range expression are registered; and a range extracting means that determines the type of the range expression by referring to the range expression table and the text. A numerical information extracting apparatus characterized in that the numerical information extracting apparatus is configured to include at least information considering a type of a range expression to the numerical information. 8. The information processing apparatus according to claim 7, further comprising an information selection unit for selecting the numerical information according to a type of a range expression included in the numerical information before outputting the numerical information to the output unit. Numeric information extraction device. 9. An omission interpolation means for referring to the text and the numerical information to correct the numerical information by supplementing the omitted character string with the numerical information in which a part of a character string representing a numerical value is omitted. The numerical information extracting apparatus according to any one of claims 1 to 8, further comprising: 10. Referring to the text and the numerical information, supplementing the omitted unit with the omitted numerical information,
9. The numerical information extracting apparatus according to claim 1, further comprising an omission interpolation unit for correcting the numerical information. 11. A numerical information in which a part of a character string representing a numerical value or a unit is omitted by referring to the text and the numerical information, and one or both of the omitted character string and the unit are supplemented. 9. The numerical information extracting apparatus according to claim 1, further comprising an omission interpolation unit for correcting the numerical information. 12. Numerical information registering means for extracting and registering numerical information from a document, condition inputting means for inputting a condition for searching for the registered numerical information, searching means for searching numerical information in accordance with the condition, a display means for outputting a search result by the search means, at least equipped with constituting, the numeric registration means, 曖numbers to the extracted numerical data representation
When a plurality of interpretations generated by昧is, numerical information retrieval apparatus characterized by registering the single numeric representation as a plurality of numerical information. 13. Numerical value registering means for extracting and registering numerical information from a document, condition inputting means for inputting a condition for searching for the registered numerical information, searching means for searching numerical information in accordance with the condition, And display means for outputting a search result by the search means. more numerical strings removed and numerical information retrieval device comprising a registration child a single numeric representation as a plurality of numeric representation based. 14. Numerical information registering means for extracting and registering numerical information from a document, condition inputting means for inputting a condition for searching for the registered numerical information, searching means for searching numerical information in accordance with the condition, Display means for outputting a search result by the search means, wherein the numerical value registration means determines a type of the unit, a unit table in which the name of the unit and the type of the unit are registered, and the document. If a plurality of interpretations occur in the type of unit as a result of referring to, a single numerical expression is registered as a plurality of numerical expressions. 15. Numerical value registration means for extracting and registering numerical information from a document, condition input means for inputting a condition for searching for the registered numerical information, search means for searching for numerical information according to the condition, And a display unit for outputting a search result by the search unit. The numerical value registration unit supplements the omitted character string with numerical information in which a part of a character string representing a numerical value is omitted. A numerical information search device comprising at least an abbreviated interpolation means for correcting numerical information, and registering a numerical expression before and after being corrected by the abbreviated interpolation means for a single numerical expression. 16. Numerical value registration means for extracting and registering numerical information from a document, condition input means for inputting a condition for searching for the registered numerical information, search means for searching for numerical information according to the condition, Display means for outputting a search result by a search means, wherein the numerical value registration means corrects the omitted numerical unit by replacing the omitted numerical value information with the omitted numerical value information. At least,
A numerical information search device, wherein a numerical expression before and after being corrected by the omitting interpolation means is registered for a single numerical expression. 17. Numerical information registering means for extracting and registering numerical information from a document, condition inputting means for inputting a condition for searching for the registered numerical information, searching means for searching numerical information according to the condition, Display means for outputting a search result by the search means, wherein the numerical value registration means is a part of a character string representing a numerical value, or a character string omitted in numerical information in which units are omitted. Or at least an abbreviated interpolation means for correcting the numerical information by supplementing the omitted unit, and registering the numerical expressions before and after being corrected by the abbreviated interpolation means for a single numerical expression. Characteristic numerical information retrieval device. 18. The display means includes at least a reference table in which a condition of numerical information required as a search result is registered, and a classification selecting means for selecting necessary numerical information from the search result by referring to the reference table. The numerical information search device according to any one of claims 12 to 17, wherein the numerical information search device is configured. 19. The display device according to claim 19, wherein the display means includes at least a reference table in which a display method for outputting the search result and correspondence between numerical information are registered, and the search result is displayed by referring to the reference table. The numerical information search device according to any one of claims 12 to 17, wherein 20. The display device according to claim 12, wherein said display means includes at least a reference table in which a display order of search results is registered, and displays the search results by referring to said reference table. A numerical information retrieval device. 21. A system according to claim 18, further comprising a reference correcting means for correcting the reference table by a user.
21. The numerical information search device according to any one of items 20 to 20. 22. Numerical value registering means for extracting and registering numerical information from a document, condition inputting means for inputting a condition for searching for the registered numerical information, searching means for searching numerical information according to the condition, Display means for outputting a search result by the search means, wherein the numerical value registration means refers to a range expression table in which a range expression and a type of the range expression are registered, and refers to the range expression table and the text. Range extraction means for determining the type of range expression, wherein the search means is a range definition means for registering a range of numerical values corresponding to the type of range expression; and A condition conversion unit that converts a search condition input from the condition input unit into a condition specification for each range type; and an information search unit that performs a search using the condition converted by the condition conversion unit. A numerical information search device comprising: 23. A numerical value extracting function for extracting a character string representing a numerical value included in a text and generating numerical information obtained by converting the character string into a numerical value; and adding a unit to the numerical information by referring to the text. A unit determination function for performing a unit determination function that adds information indicating a range of numerical values to the numerical information or a combination of the numerical information with reference to the text; and an output function that outputs the numerical information. A storage medium storing a program to be realized in a computer, wherein the numerical value extracting function detects an ambiguity of a delimiter of a character string representing the numerical value, and converts a plurality of numerical character strings from the ambiguous numerical character string of the delimiter. A storage medium storing a numerical information extraction program characterized by including at least a numerical value determination function to be taken out. 24. A numerical value extracting function for extracting a character string representing a numerical value contained in a text and generating numerical information obtained by converting the character string into a numerical value, and adding a unit to the numerical information by referring to the text. A unit determination function for performing a unit determination function that adds information indicating a range of numerical values to the numerical information or a combination of the numerical information with reference to the text; and an output function that outputs the numerical information. A storage medium storing a program to be realized in a computer, wherein the unit determination function includes: a unit table in which a unit name and a unit type are registered; and a type condition table in which a condition for determining the unit type is registered. As a result of referring to the text and the unit table to determine the type of the unit, if there is a unit having the same name but a different type, the unit is referred to the type condition table. And a unit selection function for determining a type of the numerical information, and a unit is added to the numerical information, thereby storing a numerical information extracting program. 25. A numerical value extracting function for extracting a character string representing a numerical value included in a text and generating numerical information obtained by converting the character string into a numerical value, and adding a unit to the numerical information by referring to the text. A unit determination function for performing a unit determination function that adds information indicating a range of numerical values to the numerical information or a combination of the numerical information with reference to the text; and an output function that outputs the numerical information. A storage medium storing a program to be realized in a computer, wherein the range determination function includes: a range expression table in which a range expression and a type of the range expression are registered; and a type of the range expression by referring to the range expression table and the text. And a range extracting function for determining the range information. The numerical information extracting device adds information in consideration of the type of range expression to the numerical information. The storage medium storing the program. 26. A numerical value registration function for extracting and registering numerical information from a document, a condition input function for inputting a condition for searching for the registered numerical information, a search function for searching numerical information according to the condition, A storage medium storing a program for realizing, in a computer, a display function of outputting a search result by a search unit, wherein the numerical value registration function includes a step of converting a numerical expression into an extracted numerical information.
A storage medium storing a numerical information search program characterized in that when a plurality of interpretations occur due to ambiguity , the single numerical expression is registered as a plurality of numerical information.