CN102486767B - Method and device for labeling content - Google Patents

Abstract

The present invention provides a method for content labeling, comprising: obtaining content fragments of content documents; creating a rule template, the rule template includes a set of linear orderly rules [R _f , R _t ] from R _f to R _t ; Match the rule [R _f , R _t ] on the content data of [S _f , S _t ], identify and obtain the matched data item, mark each matched data item with the metadata tag in the matching rule, and obtain a list of mapping relationships M, the relationship list M is structured content data, wherein S _f is the beginning of the content segment, S _t is the end of the content segment, R _f is the first rule of the rule template, and R _t is the last rule of the rule template rule. The invention also provides a content tagging device. The invention improves the efficiency of content labeling.

Description

Content labeling method and device for content document

technical field

The present invention relates to the field of digital typesetting, in particular, to a content labeling method and device.

Background technique

Computer software applications can help users create various content documents. In recent years, structured data formats, including markup languages (such as: XML, etc.) Fragments are annotated to describe the application structure of the content. Based on this application structure, further management, processing, and reuse of content have become an urgent need for users.

Content documents in some business fields present a large number of regular content fragments, such as collections of papers, collections of test questions, dictionaries, and the like. Fig. 1 shows an entry (or term) of a dictionary. Can comprise a large amount of similar notes in the dictionary, and the regularity of these notes is embodied in, and each note all comprises prefix (or be called word item, prefix), phonetic symbol, paraphrase etc.

In order to convert the dictionary in Figure 1 into structured data, it is necessary to mark the prefix, phonetic symbols, and definitions of each note as metadata, that is, add metadata information to the regular content fragments of the e-book. In the prior art, content annotation is performed manually, so the operation is very cumbersome.

Contents of the invention

The present invention aims to provide a method and device for content labeling of content documents, so as to solve the problem of cumbersome manual content labeling.

In an embodiment of the present invention, a content labeling method is provided, including: obtaining a content fragment of a content document; creating a rule template, which includes a set of linearly ordered rules from R _f to R _t [R _f , R _t ]; match the rule [R _f , R _t ] on the content data of [S _f , S _t ], identify and obtain the matching data item, and mark the metadata mark in the matching rule for each matched data item , to obtain the mapping relationship list M, the relationship list M is structured content data, wherein, S _f is the beginning of the content segment, S _t is the end of the content segment, R _f is the first rule of the rule template, R _t is the last rule of the rule template; wherein, the rule includes: a conditional matching rule, a repeated matching rule, and a template reference rule; the rule includes the following attributes: metadata tag, minimum number of occurrences, and maximum number of occurrences.

In an embodiment of the present invention, a content labeling device for a content document is provided, including: an acquisition module for acquiring content fragments of the content document; a creation module for creating a rule template, the rule template including from R _f A set of linear and ordered rules [R _f , R _t ] to R _t ; matching module, used to match the rules [R _f , R _t ] on the content data of [S _f , S _t ], identify and obtain matching data Items, mark the metadata tags in the matching rules for each data item that is matched, so as to obtain the mapping relationship list M, the relationship list M is structured content data, wherein, S _f is the beginning of the content segment, S _t is the end of the content fragment, R _f is the first rule of the rule template, and R _t is the last rule of the rule template; wherein, the rules include: conditional matching rules, repeated matching rules and template reference rules; the rules include the following Attributes: Metadata Tag, Minimum Occurrences, and Maximum Occurrences.

The content tagging method and device of the embodiments of the present invention overcome the problem of cumbersome manual content tagging operations and improve the efficiency of content tagging because rules are used to automatically match content segments.

Description of drawings

The accompanying drawings described here are used to provide a further understanding of the present invention and constitute a part of the application. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations to the present invention. In the attached picture:

Figure 1 shows an entry of a dictionary;

Fig. 2 shows a flow chart of a content tagging method according to an embodiment of the present invention;

Fig. 3 shows a flow chart of matching rules [R _f , R _t ] on content data of [S _f , S _t ] according to a preferred embodiment of the present invention;

FIG. 4 shows a schematic diagram of a word rule template according to a preferred embodiment of the present invention;

Fig. 5 shows a schematic diagram of the entry metadata information obtained by performing rule matching on the entry in Fig. 1 by the entry rule template in Fig. 4;

Fig. 6 shows another entry of the dictionary;

Fig. 7 shows a schematic diagram of the entry metadata information obtained by performing rule matching on the entry in Fig. 6 by the entry rule template in Fig. 4;

Fig. 8 shows a schematic diagram of a note after content marking according to a preferred embodiment of the present invention;

Fig. 9 shows a schematic diagram of a word rule template according to a preferred embodiment of the present invention;

Fig. 10 shows a schematic diagram of the entry metadata information obtained by performing rule matching on the entry in Fig. 8 by the entry rule template in Fig. 9;

Fig. 11 shows a schematic diagram of a content tagging device according to an embodiment of the present invention.

Detailed ways

The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

Fig. 2 shows a flow chart of a content labeling method according to an embodiment of the present invention, including:

Step S10, acquiring content fragments;

Step S20, match the rule [R _f , R _t ] on the content data of [S _f , S _t ], mark each matched data item with the metadata mark in the matched rule, to obtain the mapping relationship list M, where _, S _f is the beginning of the content fragment, S _t is _the end of the content fragment, R _f is the first rule of the rule template, R _t is the last rule of the rule template, and the rule template includes a set of linear orderly rules.

The existing technology uses manual content tagging, so the operation is very cumbersome. In this embodiment, the rules are pre-built, and the rules are used to match the content fragments, so as to automatically match each data item, and the pre-created rules in the rules Metadata tags are automatically matched to each data item, and through the creation of rules, these operations can be realized by computers, which improves the efficiency of content tagging.

In addition, in this embodiment, the rule [R _f , R _t ] is a set of linearly ordered rules. This rule template has a simple structure, and users can easily create this rule template for content documents of various business types. Moreover, the computer executes the linear sorting rules to match one by one, the algorithm is simple to implement and the efficiency is high.

Fig. 3 shows a flow chart of matching rules [R _{f , R t ] on content data of [S f , S t} ] according to a preferred embodiment of the present invention, including:

1. Set the current rule R _c to R _f ;

2. Perform R _c matching with S _f as the starting point to obtain the data item matched by R _c , the success flag, and the end position S _r , mark the metadata tag in R _c for the data item matched by R _c , and obtain the mapping relationship list M _r ;

3. Determine whether the success flag is valid;

4. If yes, add M _r to M, otherwise end the process;

5. Determine whether R _c is R _t , if so, end the process;

6. Otherwise, judge whether S _r is S _t , and if so, end the processing;

7. Otherwise, set S _f to S _r , set R _f to the next rule of R _c , and return to step 1.

Utilizing that [R _f , R _t ] is a set of linearly sorted rules, this preferred embodiment designs such a cyclical traversal process, which can automatically sequence all the rules of [R _f , R _t ] to the content segment [S _f , S _t ] content data to complete the matching. The process is simple and easily implemented by computer.

Preferably, R _c includes a data matching condition, and performing R _c matching includes: using the data matching condition to match each data item on the content data of [S _f , S _t ], and setting a success flag accordingly. This preferred embodiment provides a conditional matching rule, which can identify data items in the content segment by way of conditional judgment.

Preferably, R _c also includes a termination position flag, the termination position flag is invalid, used to indicate that the data matching condition is an interval condition; the termination position flag is valid, used to indicate that the data matching condition is a position condition, and the interval condition is used to indicate that the data matching condition is a position condition, and the interval condition is used to indicate that the data matching condition is a position condition. Format rules for data on a continuous interval, where the corresponding data item is data in a continuous range that meets the format rules starting from the end position of the previous data item; the position condition is used to indicate the format rule for the data set at the end position , where the corresponding data item is the data between the end position of the previous data item as the starting point and the position satisfying the format rule as the end point, where the format rule is used to indicate the regularity characteristics exhibited by the data. This preferred embodiment provides interval conditions and location conditions for the condition matching rules. When the user can determine the characteristics of the business content on a continuous interval, the interval condition can be used to achieve matching. When the user can determine that the business content is in a certain position When the characteristics of , you can use the location condition to achieve matching. This preferred embodiment can meet the content labeling requirements of various types of business content.

Preferably, the format rules include at least one of the following: content format rules, display format rules, markup format rules and Any rules, the content format rules are used to indicate the regular characteristics of data displayed on the document content; the display format rules are used to indicate The regularity of the data in the layout presentation; the mark format rules are used to indicate the regularity of the data in the application logic; the Any rule is used to indicate that any data meets the matching conditions. On the basis of the above-mentioned preferred embodiments, this preferred embodiment further indicates a variety of format rules, so as to better meet the content labeling requirements of various types of business content.

Preferably, R _c includes a repetition rule number, which is used to indicate the repeated application of the number of repetition rules in [R _f , R _t ]. This preferred embodiment provides a repeated matching rule. For example, for a note, since a note usually contains only one prefix, it is obviously not necessary to use repeated matching rules for matching for the identification of the prefix. In addition, a word entry may contain multiple meanings, so it is more appropriate to use the repeated matching rule of this preferred embodiment for identification.

Preferably, the rules include: the minimum number of occurrences, whose value is N, is used to indicate that the data items are matched at least N times, and N is a non-negative integer; the maximum number of occurrences, whose value is P, is used to indicate that the most data items are matched P times, P is a positive integer, and when N is 0, P>N; when N is a positive integer, P≥N.

Preferably, the content labeling method further includes: traversing each mapping relationship in M, respectively recording each metadata tag and corresponding data item to construct a metadata item; constructing a metadata item table from the metadata item; Tables are appended to content fragments.

Preferably, the content labeling method further includes: traversing each mapping relationship in M, respectively recording each metadata tag and corresponding data item to construct a metadata item; according to the continuous interval or end position corresponding to the matching data item, the Metadata items are attached to content fragments.

The above two preferred embodiments provide two simple and feasible schemes for saving the established mapping relationship list.

Preferably, the metadata tag conforms to XML, wherein, when the metadata tag is an empty tag, it is used to indicate that a data item whose metadata tag is an empty tag is ignored when appending the metadata item. XML is a relatively common computer language in the industry at present, and the use of XML to specify metadata tags can improve the generality of this method. In addition, by providing empty tags, unrecognizable data content in content fragments can be processed, thereby improving compatibility with content documents.

Preferably, the content tagging method further includes: analyzing the expression law of each content segment in the content document; creating a rule template according to the expression law, and the rule template includes a rule [R _f , R _t ]. By pre-creating rule templates, when identifying multiple electronic documents with similar expressions, a rule template can be shared, which avoids the need to re-establish the rules [R _f , R _t ] each time, thereby improving the reusability of content labeling.

Preferably, Rc includes a reference template name, which is used to indicate reference to a rule template with the template name. In this preferred embodiment, template reference rules are established, which can reduce the development workload.

A rule template contains a set of linear and orderly rules, which can be assigned a name. 1) The rule template can be stored and used on other similar content fragments, which is similar to the style comparison; 2) Other rule templates can also be referenced by this name Defined rule templates.

Fig. 4 shows a schematic diagram of a rule template of a note according to a preferred embodiment of the present invention. The rule template "note" contains 6 matching rules in a linear order. FIG. 5 shows a schematic diagram of entry metadata information obtained by performing rule matching on the entry in FIG. 1 by the entry rule template in FIG. 4 .

In this preferred embodiment, the rules are divided into three categories: conditional matching rules, repeated matching rules and template reference rules. Either rule contains the following properties:

Wherein, any rule can specify the minimum number of occurrences and the maximum number of occurrences, and the minimum number of occurrences <= the number of matching items <= the maximum number of occurrences can be regarded as a successful match.

Example: Answers to multiple-choice questions may appear in some problem sets as formatted text like this:

Answer: AC

It is possible to identify each alternative answer (“answer option” = “A”, “answer option” = “B”) through the continuous conditional rule ({capital letter, answer option, 1..*}).

Conditional rules can be further subdivided into two categories: continuous conditional rules and end conditional rules.

Successive conditional rules all contain the following properties:

Attributes illustrate format rules (conditions) Specifies the format rule for the match. (continuous range)

End condition rules all contain the following properties:

Among them, the mark containing the end position is not a mark to distinguish the continuous condition rule from the end condition rule, it indicates whether the scope of the matching item contains the data of the end position.

Take the stroke rules in the example as an example:

When "include end position" is TRUE ({text: "painting,", TRUE, stroke, 1}), the matching item is recognized as "4 painting," and after the end position is ",", contains "painting," ;

When the "include end position" is FALSE ({text: "painting,", FALSE, stroke, 1}), it is recognized that the matching item is "4", and after the end position is "4", "painting," is not included.

Recurrence rules all contain the following properties:

Attributes illustrate Number of repeating rules Specifies that several matching rules are applied repeatedly.

Template reference rules all contain the following properties:

Attributes illustrate reference template name Specifies to apply a defined rule template.

Wherein, the template reference rule specifies to apply the rule interval [R _f-template , R _t-template ] identified by the reference template name, which is a method of nested application.

For example: take the rule template "note" in the example as an example, change the last paraphrase rule to a template reference rule ({template reference: "paraphrase", paraphrase, 1}), when applying the paraphrase rule, the rule template will be found automatically The rule interval corresponding to "interpretation" (as shown in Figure 9) is matched.

The effect of metadata is as follows:

<metadata>

<word>open</word>

<traditional font>open</traditional font>

<Pinyin>kāi</Pinyin>

<stroke>4</stroke>

<radical>one</radical>

<interpretation>

  <meaning item> to open: ~door|~curtain|public~|network~side. </meaning item>

<meaning item> get through; open up: ~ road | ~ mine | ~ dig | ~ extension. </meaning item>

</interpretation>

</metadata>

The specific matching process is as follows:

(1) observe and analyze the data of the content segment, find its expression law, and create the rule template "note" as shown in Figure 4;

(2) select the content fragment of the note "open";

(2) The rule template contains a set of ordered matching rules;

(3) Analyze content fragments, identify and obtain matching data items according to the rule template "note", and map to associated metadata tags.

(4) Establish metadata information as shown in FIG. 5 according to the identified matching data items and associated metadata tags. These metadata information can be attached as a whole to the content fragment of the note "on".

The above step (3) can be further refined, and the rule template identification and matching includes the following steps:

(3.1) starting position S _f is set to the beginning of the content fragment (" opening " word at the beginning of the paragraph), and the end position S _t is set to be the end of the content fragment (period at the end of the paragraph);

(3.2) Set the start rule R _f as the first rule of the rule template (rule 1 "words"), and set the end rule R _t as the last rule of the rule template (rule 6 "interpretation").

(3.3) On the content data of the interval [S _f , S _t ], execute the rule matching of the interval [R _f , R _t ] to obtain the mapping relationship list M.

The above step (3.3) can be further refined, including the following steps:

(3.3.1) Set the current rule R _c as the initial rule R _f (rule 1 "character");

(3.3.2) With the starting position S _f (the "open" word at the beginning of the paragraph) as the starting point, on the content data of the interval [S _f , S _t ], execute the matching of the rule R _c ; obtain the matching of the rule R _c Success flag (effective), matching end position S _r ("(" behind the "open" word) and matching mapping relationship list M _r ("word"="open");

(3.3.3) If the successful flag matched by the rule R _c obtained in step (3.3.2) is valid, record the matching mapping relationship list M _r in the mapping relationship list M;

(3.3.4) judge whether need to continue to match (require), if yes, then enter step (3.3.5); Otherwise, terminate processing;

(3.3.5) Set the start position S _f as the end position S _r ("(" behind the "open" word) matched by the rule R _c obtained in step (3.3.2); the start rule R _f will be set as the current The next rule of the rule R _c (rule 2 "traditional characters"); go to step (3.3.1), on the content data of the interval [S _f , S _t ], execute the rule of the interval [R _f , R _t ] match.

The above steps (3.3.2) can be further refined, and the matching of a single rule includes the following steps:

(3.3.2.1) If the current rule R _c is a conditional matching rule, then enter step (3.3.2.2); if the current rule R _c is a repeated matching rule, go to step (3.3.2.4); otherwise, the current rule Rc is a template reference Matching rules, go to step (3.3.2.5);

(3.3.2.2) According to the conditions in the current conditional rule _Rc , including the termination position mark and the number of occurrences, obtain the matching data item list and the success mark that satisfy the rule;

(3.3.2.3) For each matching data item in the matching data item list obtained in step (3.3.2.2), establish a mapping relationship with the metadata tag in the current conditional rule R _c , and add the mapping relationship list M _r In; terminate processing.

(3.3.2.4) According to the number of repetition rules and the number of occurrences in the current repetition rule R _c , on the content data of the interval [S _f , S _t ], repeatedly execute the interval [R _{c-number of repetitions} , R _c-1 ] The rules are matched, and the mapping relationship list M _r is obtained; the processing is terminated.

(3.3.2.5) According to the reference template name and the number of occurrences in the current template reference rule R _c , on the content data of the interval [S _f , S _t ], execute the interval identified by the reference template name [R _f-template , R _t-template ] to match the rules to obtain the mapping relationship list M _r ; terminate the processing. Take rule 1 "words" as an example, this rule is a continuous conditional rule (that is, an interval condition is used). All content starting from "Kai" with a font size of No. 3 satisfies this condition, so its matching data item is "Kai" at the beginning of the paragraph.

Taking the rule 4 "stroke" as an example, this rule is an end condition rule (that is, a position condition is adopted). All the content starting from the back of the pinyin letter and ending with the string of "painting," all meet this condition, so the matching data item is "4hua,".

The rule template established in the above preferred embodiment is also applicable to other content segments conforming to the rule. For example, the word entry "californium", its content fragments and metadata information, as shown in Figure 6 and Figure 7, Figure 6 shows another word entry of the dictionary, Figure 7 shows the effect of the word entry rule template in Figure 4 on the word entry in Figure 6 Schematic diagram of entry metadata information obtained by rule matching.

Preferably, the content tagging method further includes: taking the matched data item as a content segment, and continuing to perform the step of matching the rule [R _f , R _t ] on the content data of [S _f , S _t ]. This preferred embodiment provides a nesting mechanism that can handle more complex content structures, thereby meeting the labeling requirements of content documents of various business types.

The metadata information established in the above step (4) can also be attached to the data interval of the content segment in an embedded manner, as shown in Figure 8, the matched metadata tags are all marked in the content segment of the note "Open". FIG. 9 shows a schematic diagram of a word rule template according to a preferred embodiment of the present invention, and FIG. 10 shows a schematic diagram of the word item metadata information obtained by performing rule matching on the word item in FIG. 8 by the word rule template in FIG. 9 .

It can be seen that the content of the note "Open" in FIG. 8 is further divided into smaller content fragments by the embedded metadata tags. For these lower-level content fragments, the method of the present invention can be continued to be used for labeling. Taking the content fragment of "Open Interpretation" as an example, its rule template and metadata information are shown in Figure 9 and Figure 10 .

Users can apply this content labeling method layer by layer and step by step, and label content documents or fragments at the maximum and minimum granularity, so as to achieve a satisfactory structural effect.

Figure 11 shows a schematic diagram of a content tagging device according to an embodiment of the present invention, including:

An acquisition module 10, configured to acquire content fragments;

The matching module 20 is used to match the rule [R _f , R _t ] on the content data of [S _f , S _t ], mark each matched data item with the metadata mark in the matched rule, so as to obtain a list of mapping relationships M, where S _f is the beginning of the content fragment, S _t is the end of the content fragment, R _f is the first rule of the rule template, R _t is the last rule of the rule template, and the rule template includes from R _f to R _t A set of linearly ordered rules.

The content labeling device improves the efficiency of content labeling.

Preferably, the matching module 20 includes:

The current setting module is used to set the current rule R _c to R _f ;

The current matching module is used to perform _Rc matching with _Sf as the starting point, so as to obtain the data item matched by _Rc , the success flag, and the end position _Sr , and mark the metadata tag in _Rc for the data item matched by _Rc , Get the mapping relationship list M _r ;

Adding a module for adding M _r to M if the success flag is valid;

Judgment module, used to judge whether the success flag is effective, and whether R _c is not R _t , and whether S _r is not S _t ;

The loop module is used to set S _f to S _r and R _f to the next rule of R _c if the above judgments are all yes, and then continue to execute the above steps; otherwise, terminate the process.

The content labeling device has a simple structure and can be easily realized by a computer.

Preferably, the content labeling device further includes: a metadata item module, configured to traverse each mapping relationship in M, and record each metadata tag and corresponding data item to construct a metadata item; a metadata item table module, used The metadata item is used to construct a metadata item table; an additional module is used for attaching the metadata item table to the content fragment. This preferred embodiment provides two simple and feasible solutions for saving the established mapping relationship list.

Preferably, the content tagging device further includes: an analysis module, for analyzing the expression law of each content segment in the content document; a creation module, for creating a rule template according to the expression law, and the rule template includes a rule [R _f , R _t ]. This preferred embodiment improves the reusability of content labeling work.

Various embodiments of the present invention can be combined with batch processing technology or macro command operations, so that batches of content segments with specific regularity can be quickly identified, matched, and metadata information added.

It can be seen from the above description that through various embodiments of the present invention, users can conveniently, flexibly, efficiently and accurately add metadata information to regular content segments. The above-mentioned embodiments of the present invention, combined with the metadata marking system, can be applied to various application fields, such as chapters, papers, test questions, word (dictionary) dictionaries, etc., to meet different business needs of users.

Obviously, those skilled in the art should understand that each module or each step of the above-mentioned present invention can be realized by a general-purpose computing device, and they can be concentrated on a single computing device, or distributed in a network formed by multiple computing devices Optionally, they can be implemented with program codes executable by computing devices, so that they can be stored in storage devices and executed by computing devices, or they can be made into individual integrated circuit modules, or their Multiple modules or steps are implemented as a single integrated circuit module. As such, the present invention is not limited to any specific combination of hardware and software.

The above descriptions are only preferred embodiments of the present invention, and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and changes. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims (17)

1. A content labeling method for a content document, characterized in that, comprising: Get the content fragment of the content document; Create a rule template, which contains a set of linearly ordered rules from R _f to R _t [R _f , R _t ]; Match the rule [R _f , R _t ] on the content data of [S _f , S _t ], identify and obtain the matched data item, mark each matched data item with the metadata tag in the matching rule, and obtain a list of mapping relationships M, the relationship list M is structured content data, wherein S _f is the beginning of the content segment, S _t is the end of the content segment, R _f is the first rule of the rule template, and R _t is the the last rule of the above rule template; Wherein, the rules include: condition matching rules, repeated matching rules and template reference rules; The rule contains the following attributes: Metadata Tag, Minimum Occurrences, and Maximum Occurrences. 2. The method according to claim 1, wherein the matching rule [R _f , R _t ] on the content data of [S _f , S _t ] comprises: Set the current rule R _c to R _f ; Perform R _c matching with S _f as the starting point to obtain the data items matched by R _c , the success flag, and the end position S _r , mark the metadata tags in R _c for the data items matched by R _c , and obtain the mapping relationship list M _r ; If the success flag is valid, add M _r to M; judging whether the success flag is valid, and whether R _c is not R _t , and whether S _r is not S _t ; If the above judgments are all yes, then set S _f to S _r , set R _f to the next rule of R _c , and then continue to execute from the step of setting the current rule R _c to R _f to the step of judging whether the success flag is valid all steps in between; otherwise terminate processing. 3. The method according to claim 2, wherein R _c includes data matching conditions, and performing R _c matching includes: using the data matching conditions to match each data on the content data of [S _f , S _t ] item, and set the success flag accordingly. 4. method according to claim 3, it is characterized in that, _Rc also comprises termination position mark, and described termination position mark is invalid, is used for indicating that described data matching condition is an interval condition; Described termination position mark is effective , used to indicate that the data matching condition is a location condition, The interval condition is used to indicate a format rule for data set on a continuous interval, wherein the corresponding data item is data within a continuous range that satisfies the format rule starting from the end position of the previous data item; The position condition is used to indicate the format rule of the data set at the end position, wherein the corresponding data item is the data between the end position of the previous data item as the start point and the position satisfying the format rule as the end point , Wherein, the format rule is used to indicate the regularity characteristic exhibited by the data. 5. The method according to claim 4, wherein the format rules include at least one of the following: content format rules, display format rules, markup format rules and Any rules, The content format rule is used to indicate the regularity characteristics of the data in the document content; The display format rule is used to indicate the regularity of the data in the layout presentation; The markup format rule is used to indicate the regularity of the data in application logic; The Any rule is used to indicate that any data satisfies the matching condition. 6 . The method according to claim 2 , wherein R _c includes a repetition rule number, and the repetition rule number is used to indicate repeated application of the number of repetition rules in [R _f , R _t ]. 7. The method according to claim 1, wherein the rules comprise: The minimum number of occurrences, whose value is N, is used to indicate that the matched data item is at least N times, and N is a non-negative integer; The maximum number of occurrences, the value of which is P, is used to indicate that the matched data item is at most P times, P is a positive integer, and when N is 0, P>N; when N is a positive integer, P≥N. 8. The method of claim 1, further comprising: Traversing each mapping relationship in M, recording each of the metadata tags and corresponding data items to construct metadata items; Constructing the metadata item into a metadata item table; The metadata item table is appended to the content fragment. 9. The method according to claim 4, further comprising: Traversing each mapping relationship in M, recording each of the metadata tags and corresponding data items to construct metadata items; The metadata item is appended to the content segment according to matching the continuous interval or the end position corresponding to the data item. 10. The method according to claim 8 or 9, wherein the metadata tag conforms to XML, wherein when the metadata tag is an empty tag, it is used to indicate that the metadata item is ignored when the metadata item is attached. A data item whose metadata tag is an empty tag. 11. The method of claim 1, further comprising: Analyze the performance rules of each content fragment in the content document; A rule template is created according to the expression rule, and the rule template includes a rule [R _f , R _t ]. 12 . The method according to claim 11 , wherein R _c includes a reference template name, which is used to indicate reference to a rule template with the template name. 13 . 13. The method according to claim 1, further comprising: taking the matched data item as a content segment, and continuing to execute the matching rule [R _f , R on the content data of [S _f , S _t ] _t ] steps. 14. A content labeling device for a content document, characterized in that it comprises: An acquisition module, used to acquire the content fragment of the content document; Create a module for creating a rule template, the rule template includes a set of linearly ordered rules [R _f , R _t ] from R _f to R _t ; The matching module is used to match the rule [R _f , R _t ] on the content data of [S _f , S _t ], identify and obtain the matched data item, and mark the metadata mark in the matching rule for each matched data item, To obtain a mapping relationship list M, the relationship list M is structured content data, wherein S _f is the beginning of the content segment, S _t is the end of the content segment, and R _f is the first rule of the rule template , R _t is the last rule of the rule template; Wherein, the rules include: condition matching rules, repeated matching rules and template reference rules; The rule contains the following attributes: Metadata Tag, Minimum Occurrences, and Maximum Occurrences. 15. The device according to claim 14, wherein the matching module comprises: The current setting module is used to set the current rule R _c to R _f ; The current matching module is used to perform _Rc matching with _Sf as the starting point, so as to obtain the data item matched by _Rc , the success flag, and the end position _Sr , and mark the metadata tag in _Rc for the data item matched by _Rc , Get the mapping relationship list M _r ; Adding a module for adding M _r to M if the success flag is valid; A judging module, configured to judge whether the success flag is valid, whether R _c is not R _t , and whether S _r is not S _t ; The loop module is used to set _Sf as _Sr if the above judgments are all yes, set _Rf as the next rule of _Rc , and then continue to execute the current setting module, current matching module, adding module and judging module in sequence The step involved; otherwise terminate processing. 16. The device of claim 14, further comprising: The metadata item module is used to traverse each mapping relationship in M, and respectively record each of the metadata tags and corresponding data items to construct metadata items; A metadata item table module, configured to construct a metadata item table from the metadata item; An appending module, configured to append the table of metadata items to the content fragment. 17. The apparatus of claim 14, further comprising: An analysis module, configured to analyze the performance rules of each content segment in the content document; The creation module creates a rule template according to the expression rule, and the rule template includes a rule [R _f , R _t ].