CN105404614A - Subject and predicate coding based text watermark embedding and extraction method - Google Patents

Abstract

The invention relates to a text watermark embedding and extraction method based on subject-predicate encoding. The embedding method includes: 1) expressing each character of the watermark information with Unicode code to form a Unicode code string; 2) detecting the sentence in the text to be embedded 3) According to the number of subject and predicate, divide the Unicode code string into several segments, each subject and predicate is coded by one segment, and given a number; 4) Store each The subject predicate, the Unicode code segment corresponding to the subject predicate, and its serial number form a codebook to complete the encoding and realize the embedding of the watermark. The extraction method includes: finding out the subject and predicate in the detected text, comparing the codebook, taking out the Unicode code segment and its number corresponding to each subject and predicate, splicing the Unicode code segments according to the numbering order, and converting the obtained Unicode code string into corresponding characters to form watermark information. This method does not make any changes to the text format and content, and has good concealment and robustness. At the same time, the algorithm structure is simple and easy to implement.

Description

A text watermark embedding and extraction method based on subject-predicate coding

technical field

The invention relates to watermark embedding and extraction technology, in particular to a text watermark embedding and extraction method based on subject-predicate coding.

Background technique

With the popularization and application of the Internet and information technology, more and more text information is published, disseminated and used in digital form. While it brings convenience to people's study, work and life, it also produces texts that are easy to be illegal. The protection of intellectual property rights of digital texts has attracted widespread attention in the industry due to problems such as copying and misappropriation. Text watermarking is a technology that has emerged in recent years to protect the intellectual property rights of digital texts. It embeds copyright identification information or identity authentication information (watermarks) in digital texts in a certain way. Extract these information to prove the copyright ownership of the text, confirm illegal copying and misappropriation, and protect the rights and interests of the text copyright owner or owner. In addition, text watermarking technology can also be used to hide and transmit secret information in text, authenticate text content, track text information, etc.

Currently, there are two main methods of text watermarking—text watermarking based on text format and text watermarking based on natural language. The watermarking technology based on the text format embeds the watermark information by making use of the characteristics of slightly changing the text format that are not easy to be detected, such as changing the line spacing, word spacing, character size, and so on. This kind of watermarking technology based on text format is simple in structure and easy to implement, but the embedded watermark may be destroyed if the format of the text is changed, so the robustness is not strong. Text watermarking technology based on natural language encodes the syntax and semantics of the text content to embed the watermark information. At present, most of the implementations are to encode the watermark information through synonym replacement and syntax transformation. Compared with text-based watermarks, natural language text watermarks have better concealment and robustness, and format changes will not affect watermarks. However, due to the complexity of the Chinese language, synonym replacement and syntactic transformation may cause ambiguity or change semantics, and it is not suitable for situations where the text content should not be changed.

Contents of the invention

The purpose of the present invention is to overcome the deficiencies of the prior art above, and provide a text watermark embedding and extraction method based on subject-predicate encoding with good concealment and robustness, specifically implemented by the following technical solutions:

The text watermark embedding method based on subject-predicate coding includes

1) Express each character of the watermark information in Unicode to form a Unicode code string.

2) Detect the subject-predicate of the sentence to be embedded in the text and store it in a set.

3) According to the number of detected subject-predicates, the Unicode code string is divided into several segments, and each subject-predicate is coded by one of the segments. Considering that changing the order of sentences in the text may make the watermark information incorrectly extracted, a number is given to the Unicode code segment corresponding to each subject and predicate, which is used to splice the Unicode code string according to the number when extracting the watermark.

4) Store each subject and predicate, the Unicode code segment corresponding to the subject and predicate, and the number corresponding to the subject and predicate in sequence to form a codebook, complete the encoding, and realize the embedding of the watermark.

The above Unicode encoding adopts UTF-16 format, and each character is a 4-digit hexadecimal number, forming a hexadecimal Unicode code string.

Detecting the subject and predicate in the text to be embedded in the step 2) includes the following steps:

A) Convert the submitted text to be embedded in the watermark into a character string;

B) Submit the string of the text to be embedded with the watermark to the language technology platform LTP for dependency syntax analysis, and obtain a string in XML format containing the dependency relationship of sentence components in the text;

C) Convert the string obtained in XML format into an XML file, perform DOM analysis on the XML file, and loop through the file to find out each sentence according to the relationship between the core relationship and the subject-predicate relationship of the sentence component relationship attribute in the XML file of the subject predicate.

A further design of the text watermark embedding method based on subject-predicate coding is that the subject-predicate, Unicode code segment, and serial number of each row in the codebook are separated by spaces.

According to the text watermark embedding method based on subject-predicate encoding, a text watermark extraction method based on subject-predicate encoding is proposed, including finding the subject-predicate in the detected text, comparing the codebook formed when embedding the watermark, Take out the Unicode code segment and number corresponding to each subject and predicate from the codebook, splice the Unicode code segment in the order of the corresponding number, obtain the Unicode code string representing the watermark information, and then convert it into the corresponding character to form the embedded watermark information .

A further design of the method for extracting text watermarks based on subject-predicate coding is that the step of extracting the Unicode code segment and its number corresponding to each subject-predicate in the detected text includes: each of the detected text to be found The subject-predicate is compared with each subject-predicate in the codebook one by one, and if they are consistent, the Unicode code segment and serial number corresponding to the subject-predicate are taken out from the codebook.

The advantages of the present invention are as follows:

The invention proposes a new text watermark embedding and extraction method, which uses the subject and predicate of the sentence in the text to encode the watermark information to embed the watermark. This method does not make any changes to the format and content of the text, and does not have any influence on the original text. The embedding of the watermark has no traces, will not be noticed and found, and has good concealment. Changing the format of the text (including changing line spacing, word spacing, changing character size, font, color, etc.), adjusting text paragraphs, and changing the order of sentences will not affect the correct extraction of watermarks, which has good robustness. At the same time, the algorithm is simple in structure and easy to implement.

detailed description

The solution of the present invention will be described in detail below.

The text watermark embedding method based on subject-predicate encoding provided in this embodiment includes the following steps: 1) each character of the watermark information is represented by Unicode encoding in UTF-16 format, and each character is a 4-digit hexadecimal number, Form a hexadecimal Unicode string. 2) Detect the subject-predicate to be embedded in the text and store it in a set. 3) According to the number of detected subject-predicates, the Unicode code string is divided into several segments, each subject-predicate is coded by one of the segments, and a number is given to the Unicode code segment corresponding to each subject-predicate, which is used to extract the watermark Concatenate Unicode code strings according to numbers. 4) Store each subject-predicate, the Unicode code segment corresponding to the subject-predicate, and its number in turn to form a codebook, complete the encoding, and realize the embedding of the watermark. Among them, the subject predicate, Unicode code segment, and number of each line in the codebook are separated by spaces.

Further, detecting the subject and predicate in the text to be embedded in step 2) includes the following steps: A) converting the text submitted to be embedded with the watermark into a character string. B) Submit the string of the text to be embedded with the watermark to the language technology platform LTP for dependency syntax analysis, and obtain a string in XML format containing the dependency relationship of sentence components in the text. C) Convert the string obtained in XML format into an XML file, perform DOM analysis on the XML file, and loop through the file to find out each sentence according to the relationship between the core relationship and the subject-predicate relationship of the sentence component relationship attribute in the XML file of the subject predicate.

The above-mentioned Language Technology Platform (Language Technology Platform, LTP) is a complete set of open online Chinese natural language processing system developed by the Social Computing and Information Retrieval Research Center of Harbin Institute of Technology for ten years, including lexical analysis (word segmentation, part-of-speech tagging and named entity recognition), syntactic analysis (dependency syntactic analysis), semantic analysis (word sense disambiguation and semantic role labeling) and six language processing functions. The platform is open to the outside world and easy to use. The system provides an application programming interface (API). Users can set API parameters according to their own application requirements, construct HTTP requests, submit text content to the system, and then obtain analysis results online. Here, the dependency syntactic analysis function of LTP is mainly used, and the text to be analyzed is submitted to the platform, and the dependency relationship between the components of the sentences in the text is obtained through the platform processing. According to the dependency relationship, the subject-predicate of the sentence is obtained through further processing. The basic process is as follows:

Convert the text content to be analyzed into a string, set the API parameters and calling method, and submit the string containing the text content to LTP for dependency syntax analysis. Obtain the processing result of LTP, and obtain a file in XML format containing the dependency relationship of sentence components in the text. This file contains nodes such as para (paragraph), sent (sentence), and word (word segmentation). Each word segment node (word) has the following attributes: id, the sequence number of the word segment in the sentence; cont, the content of the word segment; parent, the id number of the parent node of the dependency syntax analysis; relate, the corresponding relationship. To find out all the subject predicates in the text, loop through each paragraph, each sentence, and each participle in the text. When looping through the word node, if the attribute relate="HED" (HED represents the core relationship), the cont value of the word node is the predicate of this sentence, and then check whether there is such a word node in this sentence, and the value of the relate attribute of the node It is SBV (SBV means subject-predicate relationship), and its parent attribute value is equal to the id number of the predicate node. If it exists, the cont value of the word node is the subject of the sentence. Extract the subject and predicate of this sentence and store them in a set.

The specific function codes and comments for implementing the watermark embedding process are given below:

According to the requirements of the LTP system application program interface API, the text that needs to be submitted for analysis is converted into a string form, which can be realized by the System.IO.File.ReadAllText(stringpath,Encoding.Default) function of the C# language, and the corresponding program code for:

Stringtext=System.IO.File.ReadAllText(path,Encoding.Default);

Among them, path is the path of the text file to be analyzed, and text is a string containing the text content.

Set API parameters, including address urlbase for accessing LTPWeb services, API key api_key (obtained during user registration), analysis mode pattern (choose dp, dependent syntax analysis), result format type format (choose XML format), HTTP request method ( Select the GET method), etc., and submit the string (text) containing the text content to the LTP platform for dependency syntax analysis. The core code to realize this process is as follows:

stringurlbase="http://api.ltp-cloud.com/analysis/";

stringapi_key="k2r3q7tqGgWp5zBZRSnEHvNKfTRSFhjMtnHQ0QeP";

stringpattern="dp";

stringformat="xml";

stringstrParam=("api_key="+api_key+"&text="+text.ToString()

+"&pattern="+pattern+"&format="+format);

Encodingencoding=Encoding. GetEncoding("utf-8");

HttpWebRequestreq=

WebRequest.Create(urlbase+strParam)asHttpWebRequest;

req.Method="GET";

Obtain the processing result of LTP, and obtain a character string in XML format containing the dependency relationship of sentence components in the text. This process can be realized with the StreamReader class of C# language, and the corresponding program code is:

HttpWebResponsewebResponse=req.GetResponse()asHttpWebResponse;

StreamReaderstreamReader=

newStreamReader(webResponse. GetResponseStream(), encoding);

Stringresult = streamReader. ReadToEnd();

The result stored in result is the processed result.

Convert the string obtained in XML format into an XML file, perform DOM analysis on the XML file, and loop according to the relationship between HED (core relationship) and SBV (subject-verb relationship) of the relate attribute of each sentence in the dependency syntax analysis result Traverse each paragraph, each sentence, each participle, and find out the subject and predicate of each sentence. The core code to realize this process is as follows:

XmlDocumentdoc=newXmlDocument();//Convert to XML file

doc. LoadXml(result);

XmlElementroot=doc.DocumentElement;//Cycle traversal

XmlNodeList list1, list2, list3;

XmlNodelist4;

list1=root.SelectNodes("//para");

foreach(XmlNodenode1inlist1){//Loop through para nodes

list2=node1.ChildNodes;

foreach(XmlNodenode2inlist2){//Loop through the sent node

list3=node2.ChildNodes;

foreach(XmlNodenode3inlist3){//loop through the word node

if(node3.Attributes["relate"].InnerText=="HED"){//judgment predicate

list4=node3;

foreach(XmlNodenode4inlist3){

if(node4.Attributes["parent"].InnerText==list4.Attributes["id"].InnerText&&node4.Attributes["relate"].InnerText=="SBV")//judgment subject hs.Add(node4.Attributes ["cont"].InnerText+list4.Attributes["cont"].InnerText+"");

}

}

}

}

}

List<string>sbv=newList<string>();

sbv. AddRange(hs);

What is stored in the set sbv is the subject and predicate in the text.

Convert the watermark information to be embedded into a Unicode code string according to UTF-16 encoding. The corresponding implementation code is:

byte[]bts=Encoding.Unicode.GetBytes(info);

for(inti=0;i<bts. Length;i+=2)

uc+=bts[i+1].ToString("x").PadLeft(2,'0')+bts[i].ToString("x").PadLeft(2,'0');

Among them, the watermark information is stored in info, and the generated Unicode code string is stored in uc.

The above-mentioned Unicode code string is encoded with the subject-predicate in the subject-predicate set. Take out each subject and predicate in the set in turn, assign a Unicode code to it, and give a number. The subject and predicate, the Unicode code segment, and the number are separated by spaces to form a codebook. The core code to realize this process is as follows.

Code that determines the number of Unicode digits assigned to each subject-predicate:

strU_size=uc.length();//strU_size is the number of digits in the Unicode code string

sbv_size=sbv.Count;//sbv_size is the number of main predicates

count_size=strU_size/sbv_size;//count_size is the number of digits assigned to the main predicate Unicode code

Assign a Unicode code to each subject and predicate:

for(intx=0;x<sbv_size;x++){

if(x==sbv_size-1){//Assign the Unicode code for the last subject and predicate (the number of digits is different, and it is processed separately)

code_list.Add(sbv[x]+""+uc.ToString().Substring(x*count_size)+""+(x+1));

}else{//Allocate Unicode codes for the previous subject and predicate (equal distribution, the same number of digits)

if(x*count_size-1>0){

code_list.Add(sbv[x]+""+uc.ToString().Substring(x*count_size,count_size)+""+(x+1));

}else{

code_list.Add(sbv[x]+""+uc.ToString().Substring(0,x*count_size+count_size)+""+(x+1));

}

}

}

The collection code_list stores the codebook content, write it into a txt file, and then get the codebook file embedded with the watermark.

According to the above text watermark embedding method based on subject-predicate encoding, a method for extracting text watermark based on subject-predicate encoding is proposed, and its specific implementation is as follows:

When the watermark needs to be extracted, the detected text is submitted to the LTP platform for dependency syntax analysis, and the analysis result is further processed to obtain the subject and predicate in the text, which is stored in a set. The code to implement this process is the same as when embedding the watermark earlier.

Open the codebook file formed when the watermark is embedded, compare the codebook, and decode each subject and predicate in the above-mentioned set. That is, each subject-predicate in the set is compared one by one with each subject-predicate in the codebook one by one, and if the two are consistent, the Unicode code segment and its number corresponding to the subject-predicate are taken out. The obtained Unicode code segments are spliced together according to their serial numbers to obtain the Unicode code string representing the watermark information.

The code and comments of the main operations to achieve the above process are given below:

Read each line of the codebook and put it into an array. The implementation code is:

string[]lines=File.ReadAllLines(path);

Among them, path is the path of the codebook file, and lines is an array containing the content of each line of the codebook.

According to the space, the subject and predicate of each line are separated, and compared with the subjects and predicates in the detected text stored in the aforementioned collection one by one, if there is a consistency, the Unicode code segment and its number of the line are taken out and put into a collection. The implementation code is:

for(inti=0;i<sbv.Count;i++){

for(intj=0;j<lines. Length;j++){

string[]lgs=lines[j].ToString().Split(newChar[]{''},2);

if(sbv[i]==lgs[0])

st.Add(lgs[1]);}

}

st is a collection of Unicode code segments and their numbers.

According to the space, separate each Unicode code segment and its number, splice each code segment according to the sequence of numbers, and obtain the Unicode code string representing the watermark information. The implementation code is:

for(intx=0;x<st.Count;x++){

for(inty=0;y<st.Count;y++){

string[]lgs=st[y].ToString().Split(newChar[]{''},2);

if(Convert.ToInt32(lgs[1])==(x+1))

drawUc.Append(lgs[0]);

}

}

In drawUc is the Unicode code string representing the watermark information.

According to the UTF-16 encoding rules used when embedding the watermark, the above Unicode code string is converted into corresponding characters, and the embedded watermark information is obtained. The core code to realize this process is as follows:

MatchCollectionmc=Regex.Matches(str,"([\w]{2})([\w]{2})",

RegexOptions.Compiled|RegexOptions.IgnoreCase);

byte[]bts=newbyte[2];

foreach(Matchminmc){

bts[0]=(byte)int.Parse(m.Groups[2].Value,NumberStyles.HexNumber);

bts[1]=(byte)int.Parse(m.Groups[1].Value,NumberStyles.HexNumber);

toStr+=Encoding.Unicode.GetString(bts);

}

The information contained in toStr is the extracted watermark information.

Claims (6)

1., based on a Text Watermarking embedding grammar for subject-predicate language coding, it is characterized in that comprising

1) by each character Unicode coded representation of watermark information, a Unicode code string is formed;

2) detect the subject-predicate language of statement in text to be embedded, deposit in a set;

3) according to the subject-predicate language quantity detected, Unicode code string is divided into some sections, each subject-predicate pragmatic wherein one section carrys out coded representation, to the given numbering of the Unicode code section that each subject-predicate language is corresponding, according to numbering splicing Unicode code string during for extracting watermark;

4) store Unicode code section corresponding to each subject-predicate language, this subject-predicate language and numbering corresponding to this subject-predicate successively, form a code book, complete coding, realize the embedding of watermark.

2. the Text Watermarking embedding grammar based on subject-predicate language coding according to claim 1, it is characterized in that described Unicode encodes and adopt UTF-16 form, each character is 4 sexadecimal numbers, forms a hexadecimal Unicode code string.

3. Text Watermarking embedding grammar according to claim 1, is characterized in that described step 2) in detect that the subject-predicate language in text to be embedded comprises the steps:

To the text-converted of watermark to be embedded be submitted to be the form of character string;

B) character string of the text of watermark to be embedded is committed to language technology platform LTP and carries out interdependent syntactic analysis, obtain the character string that comprises the XML format of sentence element dependence in text;

C) character string of the XML format obtained is converted to XML file, carries out DOM parsing to XML file, according to the contact between the Key Relationships of sentence element attribute of a relation in XML file and subject-predicate relation, searching loop file, finds out the subject-predicate language of every.

4. the Text Watermarking embedding grammar based on subject-predicate language coding according to claim 1, is characterized in that separating with space respectively between the subject-predicate language of every a line in described code book, Unicode code section, numbering.

5. the Text Watermarking embedding grammar based on subject-predicate language coding according to any one of claim 1-4, a kind of extracting method of the Text Watermarking based on subject-predicate language coding is proposed, it is characterized in that, comprise: find out the subject-predicate language in detected text, the described code book formed during contrast embed watermark, Unicode code section, numbering that each subject-predicate language is corresponding is taken out from code book, Unicode code section is got up by the sequential concatenation of the numbering of correspondence, obtain the Unicode code string representing watermark information, convert corresponding character again to, form the watermark information embedded.

6. the extracting method of the Text Watermarking based on subject-predicate language coding according to claim 5, it is characterized in that the step of Unicode code section that in described taking-up detected text, each subject-predicate language is corresponding and numbering comprises: each subject-predicate language in the detected text found out and each subject-predicate language in code book are compared one by one, if both are consistent, then from code book, take out Unicode code section, numbering that this subject-predicate language is corresponding.