CN102103612A - Information extraction method and device - Google Patents

Abstract

The embodiment of the invention discloses an information extraction method and device, which relate to the technical field of information extraction. In order to solve the problem in the prior art that the computer automatic indexing adopted cannot extract the preset text block information from the newspaper layout information and manuscript information. An information extraction method provided by an embodiment of the present invention includes: extracting text block information from a layout file, wherein the text block information includes: layout text block information and manuscript text block information; judging the text block information in the text block information whether the preset layout text block information is extracted; if the preset layout text block information is not extracted, extract the preset layout text block information; if the preset layout text block information has been extracted, then Extract the text block information of the default manuscript. By adopting the embodiments of the present invention, the workload of indexing personnel can be reduced, and the accuracy of indexing can be improved.

Description

A method and device for extracting information

technical field

The present invention relates to the technical field of information extraction, in particular to an information extraction method and device.

Background technique

With the rapid development of the Internet and information technology, digital projects in the newspaper publishing industry are also competing. In the digital information process of the newspaper publishing industry, the digital information of newspaper resources has become the core digital asset of the newspaper office. The digitized information of the newspaper resources includes: manuscript information, such as articles (text, paragraphs and titles, etc.) on the newspaper layout, text and picture content in tables, etc.; layout information, including newspaper edition, layout name, date, manuscript Location information (such as coordinate information), format information such as the font and font size of the title and text, and the association information between articles and pictures, pictures and text descriptions, etc.

In order to completely and accurately preserve the digitized information of the newspaper resources as historical data for future inquiries, or to publish cross-media in real time and accurately through a variety of digital media technologies, such as publishing through news websites, digital newspapers and CD-ROMs, etc., The layout file, that is, the digitized information of the newspaper resource, can be deciphered from the newspaper layout information by indexing software; then, the decompiled newspaper digitized information can be indexed, modified and proofread.

However, in the process of realizing the present invention, the inventor found that there are at least the following problems in the prior art: the computer automatic indexing adopted in the prior art cannot obtain information from the layout text block information and the manuscript text block information of the newspaper. Extract the preset text block information, such as: the name of the proofreader, the name of the layout designer, the name of the author, the name of the editor and other data information, so that the indexer needs to manually index one by one, which makes the workload of the indexer larger and less accurate.

Contents of the invention

An embodiment of the present invention provides an information extraction method and device to automatically extract preset text block information from the newspaper's layout text block information and manuscript text block information.

In order to achieve the above object, embodiments of the present invention adopt the following technical solutions:

On the one hand, an embodiment of the present invention provides an information extraction method, including:

Extracting text block information from the layout file, wherein the text block information includes: layout text block information and manuscript text block information;

judging whether the preset layout text block information in the text block information is extracted;

If the preset layout text block information has not been extracted, then extract the preset layout text block information;

If the preset text block information of the layout has been extracted, the preset manuscript text block information is extracted.

On the other hand, an embodiment of the present invention provides an information extraction device, including:

A text block information extraction unit, configured to extract text block information from the layout file, wherein the text block information includes: layout text block information and manuscript text block information;

a judging unit, configured to judge whether the preset layout text block information in the text block information is extracted;

A default layout extracting unit, configured to extract the preset layout text block information if the preset layout text block information has not been extracted;

The preset manuscript extracting unit is configured to extract the preset manuscript text block information if the preset layout text block information has been extracted.

An information extraction method and device provided by an embodiment of the present invention can prevent repeated extraction of the same preset layout text block information by judging whether the preset layout text block information in the text block information is extracted; if the The preset layout text block information is not extracted, then extract the preset layout text block information, thereby realizing the automatic extraction of the preset layout text block information; if the preset layout text block information has been extracted, extract The text block information of the manuscript is preset, thereby realizing the automatic extraction of the text block information of the preset manuscript.

Description of drawings

FIG. 1 is a flowchart of an information extraction method provided by an embodiment of the present invention;

Fig. 2 is a specific implementation flowchart of an information extraction method provided by an embodiment of the present invention;

Fig. 3 is a schematic structural diagram of an information extraction device provided by an embodiment of the present invention.

Detailed ways

An information extraction method and device provided in the embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

As shown in Figure 1, it is an information extraction method provided by the embodiment of the present invention, and the specific implementation process of the method is as follows:

101: Extract text block information from the layout file, wherein the text block information includes: layout text block information and manuscript text block information; wherein, the layout file can be understood as a page of a newspaper that is reversed by an indexing software The decoded digital information. The extracting the text block information from the layout file is to extract the text block information from the digitized information of the newspaper layout.

102: Determine whether the preset layout text block information in the text block information is extracted;

103: If the preset layout text block information has not been extracted, extract the preset layout text block information;

104: If the preset layout text block information has been extracted, extract the preset manuscript text block information.

An information extraction method and device provided by an embodiment of the present invention can prevent repeated extraction of the same preset layout text block information by judging whether the preset layout text block information in the text block information is extracted; if the The preset layout text block information is not extracted, then extract the preset layout text block information, thereby realizing the automatic extraction of the preset layout text block information; if the preset layout text block information has been extracted, extract The text block information of the manuscript is preset, thereby realizing the automatic extraction of the text block information of the preset manuscript.

Based on the above embodiments, as shown in FIG. 2 , it is a specific implementation flowchart of an information extraction method provided by an embodiment of the present invention. When it is necessary to extract certain preset layout text block information and preset manuscript text block information, the following process is required:

201: Set the regular expression matching rules of the preset layout text block information, the regular expression matching rules of the preset manuscript text block information, and the feature information of the preset manuscript text block information; wherein, the preset The regular expression matching rules of the layout text block information and the regular expression matching rules of the preset manuscript text block information can be expressed in the form of regular expressions; the feature information of the preset manuscript text block information can be Including: font information and location information. The regular expression matching rule of the preset layout text block information can extract the preset layout text block information from the text block information; the regular expression matching rule of the preset manuscript text block information can be extracted from the text The preset manuscript text block information is extracted from the block information; in order to obtain the preset manuscript text block information more accurately, the preset manuscript text can be obtained by narrowing down the feature information of the preset manuscript text block information The matching range of the block information, and then match the text block information of the preset manuscript within the range.

202: Extract text block information from the layout file, wherein the text block information includes: layout text block information and manuscript text block information;

203: Determine whether the preset layout text block information in the text block information is extracted;

204: If the preset layout text block information has not been extracted, extract the preset layout text block information; the specific implementation process is as follows:

S11: If the preset layout text block information has not been extracted, obtain the regular expression matching rule of the preset layout text block information; according to the regular expression matching rule of the preset layout text block information from The preset layout text block information is extracted from the layout text block information; wherein, the preset layout text block information may be the name of the editor, the name of the proofreader, the name of the layout designer, etc. in the layout information; The regular expression matching rules of the preset layout text block information can be set according to the specific needs of the layout text block information to be extracted.

S12: Set the extraction flag of the preset layout text block information to an extracted state.

It should be noted that, in order to ensure the accuracy of the extraction of the preset layout text block information, the following operations may also be performed on the extracted preset layout text block information.

S13: Verify the preset layout text block information, and give the verification result; the specific verification process is: set the preset layout text block information as the edit name in the layout text block information; Matching the extracted editing name with the name in the pre-stored editing name library, if the editing name exists in the editing name library, it is considered that the extracted preset layout text block information is correct, that is, verifying The result is 100% correct; if the extracted edit name partially matches the name in the pre-stored edit name library, or does not match at all, then the correct rate is given according to the matching status, that is, the verification result is 50% correct, Or 0% correct.

S14: According to the verification result, identify the verified text block information of the preset layout. For example: 100% correct preset layout text block information is marked as white; 50% correct preset layout text block information is marked as yellow; 0% correct preset layout text block information is marked as red.

205: If the preset layout text block information has been extracted, then extract the preset manuscript text block information; the specific implementation process may be:

If the preset layout text block information has been extracted, obtain the regular expression matching rule of the preset manuscript text block information; according to the regular expression matching rule of the preset manuscript text block information from the layout text The preset manuscript text block information is extracted from the block information.

In order to more accurately extract the text block information of the preset layout, the process of extracting the text block information of the preset manuscript in the embodiment of the present invention can also be realized through the following process: the following extracted text block information of the preset manuscript is the name of the author;

S21: When the feature information of the preset manuscript text block information includes font information, if the preset layout text block information has been extracted, obtain the preset text block information according to the font information of the preset manuscript text block information. Set the collection of manuscript text block information. For example: set the font information as: black body; then if the text block information of the preset layout has been extracted, all the text block information in the text block information of the manuscript whose font is bold is extracted, and the extracted The combination of information is the preset manuscript text block information set {T}.

In order to obtain the preset manuscript text block information more accurately, the embodiment of the present invention can further narrow the scope of obtaining the preset manuscript text block information by setting the feature information to include: position information; After the manuscript text block information collection {T}, proceed as follows:

S22: When the feature information of the preset manuscript text block information also includes location information, perform preprocessing on the preset manuscript text block information set, and respectively obtain the preset manuscript text block information set {Ts} And the preset manuscript text block information set {Te}; for example: set the position information as: the beginning of the preset manuscript text block information set content to the position Ps where the first reference character appears; and/or, the Preset the position Pe from the end of the content of the manuscript text block information set to the last occurrence of the reference symbol.

The process of preprocessing the preset manuscript text block information set {T} may specifically include: the content T to be extracted may be caused by inconsistent font descriptions in the preset manuscript text block information set {T} There is an issue with inconsistent parentheses.

S23: According to the location information, extract the subset {A} of the preset manuscript text block information from the preset manuscript text block information set {T}; Ps, extract the corresponding information a1 from the preset manuscript text block information set {Ts}, if a1 is extracted, use a1 as the subset {A}; if a1 is not extracted, then follow the position information Pe , extract corresponding information a2 from the preset manuscript text block information set {Te}, and use a2 as a subset {A}.

S24: Extract the preset manuscript text block information from the subset of the preset manuscript text block information according to the set regular expression matching rule of the preset manuscript text block information; set the preset manuscript text block information The number of matching levels of the regular expression matching rules of information is 4; wherein, the number of regular expression matching rules of the matching level 1 is 3, the number of regular expression matching rules of the matching level 2 is 3, and the matching level The number of regular expression matching rules of 3 is 2, the number of regular expression matching rules of the matching level 4 is 1; the separator is a comma or semicolon; the regular expression matching rules of each matching level form a matching set; This step can specifically include:

A regular expression matching rule corresponding to the matching level is sequentially obtained from the matching set according to the matching level; the description mode of the regular expression matching rule is a regular expression. The steps are specifically:

First, obtain the three regular expression matching rules corresponding to matching level 1 from the matching set; the rules are as follows:

Rule 1 can be: ∧(.*?(Reporter|Reporter Group|Author|Intern|Correspondent|Text∨Photo|∨Photo|Text∨Graph|Illustration|Caricature|Drawing|Practice Reporter|∨Text|Reviewer|Comment ).*?\)/g;

The above regular expression means that the full text matches "(", and matches "non-carriage return" zero to unlimited times, and matches "reporter" or "reporter group" or "author" or "intern" or "correspondent" or "text /photo” or “/photo” or “text/graph” or “illustration” or “caricature” or “drawing” or “intern reporter” or “/text” or “commentator” or “comment”, and match “non Carriage return" zero to unlimited times, and matches ")".

Rule 2 can be: ∧(\s*([\u4e00-\u9fa5]{2, 5}\s+[\u4e00-\u9fa5]{2, 5}\s*)+\s*\)/g;

The above regular expression means that the full text matches "(", and matches "blank characters" zero to unlimited times, and matches 2 to 5 Chinese characters, and matches a "blank character", and matches 2 to 5 Chinese characters, And matches "whitespace character" zero to infinite times, and matches "whitespace character" zero to infinite number of times, and matches ")".

Rule 3 can be: /(reporter|reporter|author|intern|correspondent|intern reporter|commentator|drawing|comic|illustration|writing)(:|\s|∨)*[\u4e00-\u9fa5] {2, 6}\s*(?＝($|\n|∨Photo|∨Text|From|Comprehensive Report|Text∨Photo|∨Painting|Text and Photo|Photography|∨Drawing|Organization|Excerpt Integration |Photography|[\u4e00-\u9fa5]{2, 5}Special Telegraph|Photography|Text∨Graph|Report|Written|\(This Newspaper[\u4e00-\u9fa5]*electricity\)|This edition [\u4e00-\ u9fa5]*))/g;

The above regular expression "(reporter|reporter group|author|intern|correspondent|intern reporter|commentator|drawing|caricature|illustration|writing)" means to match "reporter" or "reporter group" or "author" or " intern" or "correspondent" or "trainee reporter" or "reviewer" or "cartographer" or "cartoon" or "illustrator" or "writing";

The above regular expression "(:|\s|∨)*[\u4e00-\u9fa5]{2, 6}\s*" means to match ":" or "blank character" or "/" zero to infinite times, and Match 2 to 6 Chinese characters, and match "blank characters" zero to unlimited times;

The above regular expression "(?=" indicates the suffix of the text to be matched by the assertion;

The above regular expression "($|\n|∨Photo|∨Text|From|Comprehensive Report|Text∨Photo|∨Painting|Text and Photo|Photography|∨Drawing|Organization|Excerpt Integration|Photo|[\u4e00 -\u9fa5]{2,5}Special Electric|Photography|Text∨Graph|Report|Writing|\(This Newspaper[\u4e00-\u9fa5]*electricity\)|This edition[\u4e00-\u9fa5]*)" is Suffix content, that is, the matching position is followed by the end of the string or a "carriage return", or any of the following strings: "/photo", "/text", "sent from", "comprehensive report", " Text/Photography,/Painting, Text and Photo, Photographic Reporting,/Drawing, Arranging, Excerpt Integration, Photographing, Photography, Text/Picture, Reporting "采写" or match "this newspaper" followed by more than one Chinese character followed by "dian" or match "this edition" followed by more than one Chinese character;

The above "/g" indicates all matching characters that appear in the full-text search.

Secondly, three regular expression matching rules corresponding to matching level 2 are obtained from the matching set; the rules are as follows:

Rule 1: /(\.|,|\?|!|^|\r|\n|∨Photo|Text∨Photo|∨Painting|Text Combined|∨Drawing|Text∨Graph|∨Text Arrangement|∨Practice Health|∨text)\s*[\u4e00-\u9fa5]{2, 4}\s*(?＝((photo[\n$])|∨text|text∨photo|∨painting|text and photo| ∨Drawing|Photography|Text∨Graph|∨Text Arrangement|∨Intern))/g;

The above regular expression "(\.|,|\?|!|^|\r|\n|∨photo|text∨photo|∨painting|text and photo|∨drawing|text∨graph|∨text arrangement|∨ Intern|∨文)" means matching "." or "," or "?" or "!" or "\r" or "\n" or "/photo" or "text/photo" or "/painting" Or "text and photo" or "/drawing" or "text/graph" or "/text arrangement" or "/intern" or "/text";

The above regular expression "s*[\u4e00-\u9fa5]{2, 4}\s*" means matching "blank characters" zero to unlimited times, and matching 2 to 4 Chinese characters, and matching "blank characters" zero to infinite times;

The above regular expression "(?=" indicates the suffix of the text to be matched by the assertion;

The above regular expression "((photo[\n$])|∨text|text∨photograph|∨painting|text and photo|∨drawing|photography|text∨graph|∨text arrangement|∨intern)" indicates the content of the suffix , which matches "/photo" followed by a newline or the end of a string or "/文" or "文/照" or "/攻略" or "文典典" or "/drawing" or "photography" or "文/图" or "/Text finishing" or "/Intern"; the last ")" indicates the end of the suffix;

The above "/g" indicates all matching characters that appear in the full-text search.

Rule 2: /(reporter|reporter|author|intern|correspondent|intern reporter|commentator)(\s*|∨)+[\u4e00-\u9fa5]{2, 4}(\s+[\u4e00- \u9fa5]{2,6}){1,}\s*(?＝($|\n|∨Photo|From|Comprehensive Report|Text∨Photo|∨Painting|Text and Photo|Photo Report|Report Photography |∨Drawing|Sorting|Shooting|[\u4e00-\u9fa5]{2, 5}Special Electric|Photography|Text∨Graph|Report|Written|))/g;

The above regular expression "(journalist|journalist|author|intern|correspondent|intern|commentator)" means: match the string "journalist" or "journalist" or "author" or "intern" or " correspondent" or "trainee reporter" or "commentator";

The above regular expression "(\s*|∨)" means matching "blank character" zero or more times, or matching "/"; where "+" means and matching "(\s*|∨)" more than once ;

The above regular expression “[\u4e00-\u9fa5]{2, 4}” means matching 2 to 4 Chinese characters;

The above regular expression "(\s+[\u4e00-\u9fa5]{2, 6}) {1, }" is expressed as follows, "(\s+[\u4e00-\u9fa5]{2, 6}" means repeated matching" Blank character" more than once, match 2 up to 6 Chinese characters. "{1,}" means repeat matching "(\s+[\u4e00-\u9fa5]{2,6}" more than once;

The above regular expression "\s*" means repeat matching "blank character" zero or more times;

The above regular expression "(?=" indicates the suffix of the text to be matched by the assertion;

The above regular expression "($|\n|∨Photo|From|Comprehensive Report|Text∨Photo|∨Painting|Text and Photo|Photography|Report Photography|∨Drawing|Organization|Photo|[\u4e00-\u9fa5 ]{2,5}Special Telegraph|Photo|Text∨Graph|Report|Writing|)" is the suffix content mentioned above, indicating that the matching position is followed by the end of the string or carriage return or "/Photo" or "From" Or "comprehensive report" or "text/photo" or "/painting" or "text and photo" or "photographic report" or "road photography" or "/drawing" or "sorting" or "photographing" or 2 to 5 Chinese characters followed by "special electricity" or "photography" or "text/picture" or "report |" or "collection"; the last ")" means the end of the suffix;

The above "/g" indicates all matching characters that appear in the full-text search.

Rule 3: /(◆|●|□|◎)\s*.*(?=($|\r|\n))/g;

The above regular expression "(◆|●|□|◎)\" indicates that the string matches "◆" or "●" or "□" or "◎";

The above regular expression "\s*.*" means zero or more times of repeated matching of blank characters, and zero or more times of repeated matching of non-newline characters;

The above regular expression "(?=" indicates the suffix of the text to be matched by the assertion;

The above regular expression "($|\r|\n)" indicates the content of the suffix, and the ")" immediately following the end of the matching string or carriage return and line feed indicates the end of the suffix;

The above "/g" indicates all matching characters that appear in the full-text search.

Again, the two regular expression matching rules corresponding to matching level 3 are obtained from the matching set; the rules are as follows:

Rule 1: ∧(\s*[\u4e00-\u9fa5]{2, 4}(\s+[\u4e00-\u9fa5]{2, 6}){1,}\s*\)/g;

The above regular expression "\s*[\u4e00-\u9fa5]{2, 4}" means repeating matching "blank character" zero or more times, matching 2 to 4 Chinese characters;

The above regular expression "(\s+[\u4e00-\u9fa5]{2, 6})" means matching "blank character" more than once, matching 2 to 6 Chinese characters;

The above regular expression "{1,}" means to match "(\s+[\u4e00-\u9fa5]{2, 6})" more than once;

The above regular expression "\s*" means repeat matching "blank character" zero or more times;

The above "/g" means all matching characters that appear in the full-text search;

Rule 2: re=∧(\s*[\u4e00-\u9fa5]{2, 4}\s*\)/g;

The above regular expression means to repeatedly match "blank characters" zero or more times, match 2 to 4 Chinese characters, and repeatedly match "blank characters" zero or more times;

Among them, "/g" means all matching characters that appear in the full-text search;

Finally, a regular expression matching rule corresponding to matching level 4 is obtained from the matching set; the rule is as follows:

Rule 1: /(\s+|^|\?|\.|!)[\u4e00-\u9fa5]{2, 4}\s*(?＝((photo[\n$])|∨photo|∨ Text | Text ∨ Photo | ∨ Painting |

The above regular expression "(\s+|^|\?|\.|!)" means matching "blank character" more than once or the beginning of the string or matching "?" or "." or "!";

The above regular expression "[\u4e00-\u9fa5]{2, 4}\s*" means to match 2 to 4 Chinese characters, and repeat the matching "blank character" zero or more times;

The above regular expression "(?=" indicates the suffix of the text to be matched by the assertion;

The above regular expression "((photo[\n$])|∨photo|∨text|text∨photo|∨painting|text and photo|\sphoto|∨drawing|photography|text∨graph|∨text arrangement|∨ Intern)" means to match "摄" followed by a newline character or the end of the string, or to match "/摄" or "/文" or "文" or "文式摄" or followed by a blank character and "Photography" or "/drawing" or "photography" or "text/picture" or "/text arrangement" or "/intern"; the last ")" means the end of the suffix;

Among them, "/g" means all matching characters that appear in the full-text search.

According to the obtained regular expression matching rules, content matching is performed on the content in the subset of the preset manuscript text block information, and a matching result is given; for example: matching according to the three regular expressions of the matching level 1 The rules are matched with the content in the subset of the preset manuscript text block information, so that "author, Wang Yi" can be extracted and added to the set {B}, and then continue to obtain the 3 regular expressions of matching level 2 The formula matching rules are matched with the content in the subset of the preset manuscript text block information, and no information is extracted; then, the two regular expression matching rules of matching level 3 and the content of the preset manuscript text block information are obtained. The content in the subset is matched, and "correspondent, Zhao Er" is extracted, and added to the set {B}; finally, the combination of a regular expression matching rule of matching level 4 and the preset manuscript text block information is obtained Match the content in the subset, extract "Editor Zhang San", and add it to the set {B}; the set {B} is {author, Wang Yi, correspondent, Zhao Er, editor Zhang San}.

When the collection {B} is obtained as {author, Wang Yi, correspondent, Zhao Er, editor Zhang San}, the matching result can also be filtered by keywords according to the corresponding filtering rules, and the author name "Wang Yi" can be obtained. Extract the name into the author set {B1}; sequentially extract the correspondent name "Zhao Er" into the correspondent name set {B2}; extract the editor name "Zhang San" into the edit name set {B3} middle. The keyword filtering process completes the keyword removal process, keywords such as "author", "editor", "correspondent" and so on.

It should be noted that, since there may be multiple results separated by specific punctuation marks (such as commas, semicolons) in the results obtained through keyword filtering, such as {Wang Yi, Zhao Er, Zhang San}, it is necessary to filter the result set Perform re-extraction. Use specific punctuation marks as separators to cut strings to obtain multiple results, such as adding "Wang Yi" to the result set {A1}; adding "Zhao Er" to the result set {A2}; adding "Zhang San" to the result set { A3},.

It should be noted that the matching level can obtain the optimal value according to the experimental statistics.

Regular expression matching rules are all expressed in the form of regular expressions, which are composed of multiple keywords. Please refer to the relevant parameter descriptions for details, or configure them according to different instances. Each regular expression matching rule corresponds to a keyword substitution rule. Multi-level rule settings can extract all authors to the greatest extent; among them, the names include reporter names, correspondent names, photographer names, editor names, intern names, text editor names, commentator names, etc.

S25: Perform information reprocessing on the subset of preset manuscript text block information; the specific implementation process of this step may include: the described result set {A1}, {A2}, {A3}...{ An} is merged into the result set {A}; then, the result set {A} is subjected to secondary filtering of keywords for deduplication and omission processing. Specifically, information items with the same content in the result set {A} are removed, and the result set {A} is filtered again by keywords.

S26: Extract the preset manuscript text block information from the subset of the pre-processed preset manuscript text block information.

It should be noted that this method also includes:

S27: Verify the text block information of the preset manuscript, and give the verification result; the specific verification process can use the pre-stored dictionary information to verify and extract the text block information of the preset manuscript, that is, the author name set{A } correct rate, the steps are as follows:

Step 1: Obtain author A in turn, compare the established author name dictionary, and check whether they all exist. If yes, the correct rate of identifying this author set {A} is 100%. For some partial matches, or no matches at all, 60% and 0 correctness rates are respectively identified for the author set {A}.

Step 2: Set up the Chinese surname dictionary with a coverage rate of 95%, and perform a secondary accuracy rate calculation on the author set whose accuracy rate is not 100%, obtain the first character of the author string, compare it with the surname dictionary, and improve if it exists Correct rate. If it does not exist, get the first two characters of the author string and compare it with the surname dictionary. If it exists, the correct rate will increase, otherwise it will decrease.

S28: According to the verification result, identify the verified text block information of the preset manuscript.

As shown in Figure 3, it is an information extraction device provided by an embodiment of the present invention, the device includes:

A text block information extraction unit 301, configured to extract text block information from the layout file, wherein the text block information includes: layout text block information and manuscript text block information;

A judging unit 302, configured to judge whether the preset layout text block information in the text block information is extracted;

A preset layout extracting unit 303, configured to extract the preset layout text block information if the preset layout text block information has not been extracted;

The preset manuscript extracting unit 304 is configured to extract the preset manuscript text block information if the preset layout text block information has been extracted.

It should be noted that the device also includes:

A setting unit, configured to set the regular expression matching rules of the preset layout text block information, the regular expression matching rules of the preset manuscript text block information, and the feature information of the preset manuscript text block information.

It should be noted that the preset layout extraction unit 303 includes:

A rule acquisition subunit, configured to acquire the regular expression matching rules of the preset layout text block information;

The preset layout extraction subunit is used to extract the preset layout text block information from the layout text block information according to the regular expression matching rule of the preset layout text block information;

The flag setting subunit is used to set the extraction flag of the preset layout text block information to an extracted state.

It should also be noted that the preset layout extraction unit 303 also includes:

A verification sub-unit is used to verify the text block information of the preset layout and give the verification result;

The identification subunit is configured to identify the verified preset layout text block information according to the verification result.

It should also be noted that the preset manuscript extracting unit 304 is also used to obtain the regular expression matching rule of the preset manuscript text block information, and according to the regular expression matching rule of the preset manuscript text block information from Extracting the preset manuscript text block information from the layout text block information; or,

When the feature information of the preset manuscript text block information includes: font information and the position information, the preset manuscript extracting unit 304 is further configured to obtain the The preset manuscript text block information set, according to the font information of the preset manuscript text block information, obtain the preset manuscript text block information set; preprocess the preset manuscript text block information set; according to the position Information, extracting a subset of the preset manuscript text block information from the preset manuscript text block information set; according to the set regular expression matching rules of the preset manuscript text block information, from the preset manuscript The text block information of the preset manuscript is extracted from the subset of text block information.

It should also be noted that the preset manuscript extraction unit 304 includes:

An information reprocessing subunit, configured to perform information reprocessing on a subset of the preset manuscript text block information;

The preset manuscript extracting subunit is configured to extract the preset manuscript text block information from the subset of the pre-processed preset manuscript text block information.

It should also be noted that the preset manuscript extracting unit 304 also includes:

A verifying subunit, used to verify the text block information of the preset manuscript, and give the verification result;

The identification subunit is configured to identify the verified text block information of the preset manuscript according to the verification result.

An information extraction method and device provided by an embodiment of the present invention can prevent repeated extraction of the same preset layout text block information by judging whether the preset layout text block information in the text block information is extracted; if the The preset layout text block information is not extracted, then extract the preset layout text block information, thereby realizing the automatic extraction of the preset layout text block information; if the preset layout text block information has been extracted, extract The text block information of the manuscript is preset, thereby realizing the automatic extraction of the text block information of the preset manuscript. Compared with the prior art, the embodiment of the present invention can not only automatically extract the preset layout text block information and the preset manuscript text block information, but also can further use the pre-stored library information and the extracted preset layout The text block information is compared with the text block information of the preset manuscript, thereby improving the accuracy of extracting the text block information of the preset layout and the text block information of the preset manuscript, thereby greatly reducing the workload of the indexing personnel and improving the extraction efficiency. Accuracy. Wherein, in the process of extracting the text block information of the preset manuscript, the present invention also narrows the range of extracting the text block information of the preset manuscript through feature information, and further improves the accuracy rate of extracting the text block information of the preset manuscript.

Through the description of the above embodiments, those of ordinary skill in the art can understand that all or part of the steps in the method of the above embodiments can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable When the program is executed, the program includes the steps of the above-mentioned method embodiment, the storage medium described, such as: ROM/RAM, magnetic disk, optical disk, etc.

The above is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Anyone skilled in the art can easily think of changes or substitutions within the technical scope disclosed in the present invention. Should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims (16)

1. An information extraction method, comprising:

extracting text block information from the layout file, wherein the text block information comprises: page text block information and manuscript text block information;

judging whether preset layout text block information in the text block information is extracted or not;

if the preset layout text block information is not extracted, extracting the preset layout text block information;

and if the preset layout text block information is extracted, extracting the preset manuscript text block information.

2. The information extraction method according to claim 1, characterized by further comprising:

and setting a regular expression matching rule of the preset layout text block information, a regular expression matching rule of the preset manuscript text block information and characteristic information of the preset manuscript text block information.

3. The information extraction method according to claim 2, wherein the step of extracting the preset layout text block information includes:

acquiring a regular expression matching rule of the preset layout text block information;

extracting the preset layout character block information from the layout character block information according to the regular expression matching rule of the preset layout character block information;

and setting the extraction identification of the preset layout text block information to be in an extracted state.

4. The information extraction method according to claim 3, wherein the step of extracting the preset layout text block information further comprises:

checking the preset layout text block information and giving a checking result;

and marking the verified preset layout text block information according to the verification result.

5. The information extraction method according to any one of claims 2 to 4, wherein the step of extracting the preset manuscript text block information comprises:

acquiring a regular expression matching rule of the preset manuscript text block information;

and extracting the preset manuscript text block information from the layout text block information according to the regular expression matching rule of the preset manuscript text block information.

6. The information extraction method according to any one of claims 2 to 4, wherein when the feature information of the preset manuscript text block information includes: during the font information, the step of extracting the preset manuscript text block information further comprises:

and acquiring the preset manuscript character block information set according to the character style information of the preset manuscript character block information.

7. The information extraction method according to claim 6, wherein when the feature information of the preset manuscript text block information further comprises: during the position information, the step of extracting the character block information of the preset manuscript further comprises the following steps:

preprocessing the preset manuscript character block information set;

extracting a subset of the preset manuscript text block information from the preset manuscript text block information set according to the position information;

and extracting the preset manuscript character block information from the subset of the preset manuscript character block information according to the regular expression matching rule of the preset manuscript character block information.

8. The information extraction method according to claim 7, wherein the step of extracting the preset manuscript text block information from the subset of the preset manuscript text block information according to a regular expression matching rule of the preset manuscript text block information comprises:

carrying out information reprocessing on the subset of the preset manuscript text block information;

and extracting the preset manuscript text block information from the reprocessed subset of the preset manuscript text block information.

9. The information extraction method according to claim 8, wherein the step of extracting the preset manuscript text block information further comprises:

checking the preset manuscript text block information and giving a checking result;

and marking the verified preset manuscript text block information according to the verification result.

10. An information extraction apparatus characterized by comprising:

a text block information extracting unit, configured to extract text block information from the layout file, where the text block information includes: page text block information and manuscript text block information;

the judging unit is used for judging whether the preset layout text block information in the text block information is extracted or not;

a preset layout extracting unit, configured to extract the preset layout text block information if the preset layout text block information is not extracted;

and the preset manuscript extracting unit is used for extracting the preset manuscript text block information if the preset layout text block information is extracted.

11. The information extraction apparatus according to claim 10, characterized by further comprising:

and the setting unit is used for setting the regular expression matching rule of the preset layout text block information, the regular expression matching rule of the preset manuscript text block information and the characteristic information of the preset manuscript text block information.

12. The information extraction apparatus according to claim 11, wherein the preset layout extraction unit includes:

a rule obtaining subunit, configured to obtain a regular expression matching rule of the preset layout text block information;

a preset layout extracting subunit, configured to extract the preset layout text block information from the layout text block information according to a regular expression matching rule of the preset layout text block information;

and the mark setting subunit is used for setting the extraction mark of the preset layout text block information to be in an extracted state.

13. The information extraction apparatus according to claim 12, wherein the preset layout extraction unit further includes:

the checking subunit is used for checking the preset layout text block information and giving a checking result;

and the identification subunit is used for identifying the verified preset layout text block information according to the verification result.

14. The information extraction apparatus according to any one of claims 11 to 13,

the preset manuscript extracting unit is also used for acquiring a regular expression matching rule of the preset manuscript text block information, and extracting the preset manuscript text block information from the layout text block information according to the regular expression matching rule of the preset manuscript text block information; or,

when the characteristic information of the preset manuscript text block information comprises: the preset manuscript extracting unit is also used for acquiring a preset manuscript block information set according to the font information of the preset manuscript block information and preprocessing the preset manuscript block information set when the font information and the position information are acquired; extracting a subset of the preset manuscript text block information from the preset manuscript text block information set according to the position information; and extracting the preset manuscript text block information from the subset of the preset manuscript text block information according to the set regular expression matching rule of the preset manuscript text block information.

15. The information extraction apparatus according to claim 14, wherein the preset manuscript extraction unit includes:

the information reprocessing subunit is used for reprocessing the information of the subset of the preset manuscript text block information;

and the preset manuscript extracting subunit is used for extracting the preset manuscript text block information from the reprocessed subset of the preset manuscript text block information.

16. The information extraction apparatus according to claim 15, wherein the preset manuscript extraction unit further comprises:

the checking subunit is used for checking the preset manuscript text block information and giving a checking result;

and the identification subunit is used for identifying the verified preset manuscript text block information according to the verification result.

Images