CN107729392B - Text structuring method, device and system and non-volatile storage medium - Google Patents
Text structuring method, device and system and non-volatile storage medium Download PDFInfo
- Publication number
- CN107729392B CN107729392B CN201710852183.3A CN201710852183A CN107729392B CN 107729392 B CN107729392 B CN 107729392B CN 201710852183 A CN201710852183 A CN 201710852183A CN 107729392 B CN107729392 B CN 107729392B
- Authority
- CN
- China
- Prior art keywords
- text
- structured
- question
- result
- structuring
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/25—Integrating or interfacing systems involving database management systems
- G06F16/258—Data format conversion from or to a database
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/10—Text processing
- G06F40/103—Formatting, i.e. changing of presentation of documents
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Databases & Information Systems (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Machine Translation (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention relates to a text structuring method, a text structuring device, a text structuring system and a nonvolatile storage medium, wherein the text structuring method comprises the steps of obtaining an unstructured text, preprocessing the unstructured text, decomposing the preprocessed unstructured text into a plurality of clauses, obtaining a question-answer database of structured entries and corresponding structured entries in the structured text, respectively matching the content of the clauses to the corresponding structured entries according to question questions in the question-answer database to obtain clause structuring results, and obtaining the structured text according to the clause structuring results.
Description
Technical Field
The invention relates to the technical field of natural language processing, in particular to a text structuring method, a text structuring device, a text structuring system and a nonvolatile storage medium.
Background
The structuralization means that information contained in a text can be decomposed into a plurality of interrelated components after being analyzed, and each component has a clear hierarchical structure. The text structuring means that an unstructured text is converted into a structured text, so that the information is expressed more objectively and vividly through a structured expression mode (such as a project, a table, a structure diagram, a flow chart and the like).
Today, in the big data era, particularly in medical technology, many free texts are generated, and the growing medical text data brings new challenges to the whole medical industry: doctors diagnose and treat patients, and a large amount of medical texts are generated in the diagnosis and treatment process. Wherein most medical text data belongs to semi-structured or unstructured data. By converting semi-structured or unstructured medical text data into structured data which can be analyzed and processed by a computer, a new breakthrough can be realized in the aspects of scientific research application, clinical diagnosis and treatment, data sharing and transmission and the like.
The traditional medical text structuring processing method is basically that doctors manually process medical text data according to medical clinical experience. However, not only is the way of structuring medical texts waste time and energy, but also the accuracy of the structuring cannot meet the expected requirements.
To achieve the problem of converting unstructured text, chinese patent document CN03124897 discloses a method and apparatus for structuring text, the method comprising the steps of, inputting structuring rules; acquiring unstructured text information; carrying out syntactic analysis on the unstructured text information to generate small text fragments; searching text segments defined in the structured rules from text units of the unstructured text information; the text fragments of the unstructured text information are structured according to the conditions determined in the structuring rules. The apparatus comprises an input device for unstructured text information; input means and storage means for structured rules; extracting means for extracting small text units from the unstructured text information; structuring means for generating structured text information according to a structuring rule; and processing means for structuring text units in the text information.
Although the method and apparatus for structuring text provided in this patent document can realize the conversion of unstructured text into structured text, it is obvious that the conversion efficiency is poor, and the conversion accuracy is not optimistic.
For another example, chinese patent application CN201610405133 discloses a text structuring method for electronic medical records, which comprises the following steps: s1, loading a medical knowledge base; s2, reading in an electronic medical record text; s3, segmenting the short sentence by utilizing a forward maximum matching algorithm to obtain words in the sentence and the part of speech and relative position relation thereof; s4, judging the semantic positive and negative of the disease information description in the short sentence; s5, extracting disease information elements; s6, repeating the steps S2 to S5 until all interested contents in the electronic medical record are obtained; s7, combining different expressions of disease information elements, combining the same disease information according to the medical synonym lexicon, and removing redundant information; and S8, storing the elements of the disease description information in a structural body/class form, and completing the structuring process.
Nevertheless, the structured method provided in the patent document can effectively extract the information related to the disease from the descriptive text of the medical record, and form the structured expression of the disease information, thereby performing deep research on the disease onset rule, the diagnosis method, the treatment effect, and the like. However, the structuring method is also poor in conversion efficiency and low in accuracy.
In summary, how to improve the conversion efficiency and accuracy of text structuring becomes one of the problems to be solved urgently by those skilled in the art.
Disclosure of Invention
In order to solve the above problems, the present invention aims to provide a text structuring method, device, system and non-volatile storage medium with good conversion efficiency and high accuracy.
In order to achieve the above object, one aspect of the present invention provides a text structuring method, wherein the method includes:
acquiring an unstructured text, preprocessing the unstructured text, and decomposing the preprocessed unstructured text into a plurality of clauses;
acquiring a structured item in a structured text and a question-answer database corresponding to the structured item;
matching the content of the clauses to corresponding structured items respectively according to question questions in a question-answer database to obtain a clause structured result;
and obtaining a structured text according to the sentence structuring result.
Further, the pre-processing comprises: the numbers and special symbols in the unstructured text are replaced by uniform symbols.
Preferably, obtaining a question-answer database of structured entries and corresponding structured entries in the structured text includes:
classifying the structured items to obtain a classification result;
and respectively setting question templates aiming at the classification results, and forming a question-answer database corresponding to the structured items according to the question templates.
Further, matching the content of the clauses to the corresponding structured entries respectively according to question questions in a question-answer database to obtain a clause structured result, including:
performing word segmentation on the sentences, inputting the obtained sentence segmentation result into a first L STM network, and performing first decoding processing on the sentence segmentation result by the first L STM network to obtain a sentence decoding result;
generating a question corresponding to a clause based on a question-answer database, performing word segmentation processing on the question, inputting an obtained question word segmentation result to a second L STM network, and performing second decoding processing on the question word segmentation result by a second L STM network to obtain a question decoding result;
and carrying out combined pairing and modeling on the first L STM network and the second L STM network according to the clause decoding result and the problem decoding result so as to obtain a clause structured result.
Preferably, according to the sentence structuring result, obtaining a structured text, including:
combining a plurality of sentence structuring results to obtain a paragraph structuring result;
and post-processing the paragraph structured result to obtain a structured text.
Further, after obtaining the structured text according to the sentence structuring result, the method further includes:
converting the structured text into vectors, storing the vectors in a result database, and performing similarity comparison on the vectors of the structured text and other vectors stored in the result database to obtain a similarity text of the structured text;
and calculating the similarity between the structured text and the similarity text.
In another aspect, the present invention further provides a text structuring apparatus, including:
the preprocessing module is used for acquiring the unstructured text, preprocessing the unstructured text and decomposing the preprocessed unstructured text into a plurality of clauses;
the item acquisition module is used for acquiring a structured item in the structured text and a question-answer database corresponding to the structured item;
the sentence structuring module is used for respectively matching the content of the sentences to the corresponding structured items according to question questions in the question and answer database so as to obtain sentence structuring results;
and the text forming module is used for obtaining a structured text according to the sentence structuring result.
And further, the preprocessing module is also used for replacing the numbers and the special symbols in the unstructured text with the uniform symbols.
Preferably, the item acquisition module includes:
the classification module is used for classifying the structured items to obtain a classification result;
and the forming module is used for setting a question template according to the classification result and forming a question-answer database corresponding to the structured items according to the question template.
Further, the sentence structuring module comprises:
the sentence segmentation processing module is used for performing word segmentation processing on the sentences and inputting the obtained sentence segmentation result into a first L STM network, and the first L STM network performs first decoding processing on the sentence segmentation result to obtain a sentence decoding result;
the question processing module is used for generating questions corresponding to the clauses based on the question-answer database, performing word segmentation processing on the questions, inputting the obtained question word segmentation results to a second L STM network, and performing second decoding processing on the question word segmentation results by a second L STM network to obtain question decoding results;
and the combined pairing module is used for carrying out combined pairing and modeling on the first L STM network and the second L STM network according to the clause decoding result and the problem decoding result to obtain a clause structured result.
Preferably, the structured text obtaining module comprises:
the merging module is used for merging a plurality of clause structured results to obtain paragraph structured results;
and the post-processing module is used for post-processing the paragraph structured result to obtain the structured text.
Further, the text structuring apparatus further includes:
the similarity judging module is used for converting the structured text into vectors and storing the vectors in the result database, and performing similarity comparison on the vectors of the structured text and other vectors stored in the result database to obtain a similarity text of the structured text;
and the similarity calculation module is used for calculating the similarity between the structured text and the similarity text.
The invention further provides a text structuring system, which comprises the text structuring device.
Still another aspect of the present invention provides a nonvolatile storage medium having a text structuring program stored thereon, the text structuring program being executed by a computer to implement a text structuring method, comprising:
the method comprises the steps of a, obtaining an unstructured text, preprocessing the unstructured text, and decomposing the preprocessed unstructured text into a plurality of clauses;
instruction b, obtaining a structured item in the structured text and a question-answer database corresponding to the structured item;
the instruction c is used for respectively matching the content of the clauses to the corresponding structured items according to question questions in the question-answer database so as to obtain a clause structured result;
and d, obtaining a structured text according to the sentence structuring result.
As described above, the text structuring method, device, system and non-volatile storage medium provided by the present invention, in combination with the question-answer database, can completely convert unstructured text information into structured information, and have good conversion effect and high accuracy, and perform sentence division structuring processing through two L STM networks, can process diverse expression modes in free text, and have good robustness.
In order to make the aforementioned and other objects of the present invention comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
The following description will explain embodiments of the present invention in further detail with reference to the accompanying drawings.
Fig. 1 is a flowchart of a text structuring method according to a first preferred embodiment of the present invention;
FIG. 2 is a diagram illustrating text of a medical record according to a first preferred embodiment of the present invention;
fig. 3 is a schematic diagram of a word segmentation result of a medical record text according to a first preferred embodiment of the present invention;
FIG. 4 is a diagram illustrating a structured result of medical record text according to a first preferred embodiment of the present invention;
FIG. 5 is a flowchart of a text structuring method according to a second preferred embodiment of the present invention;
FIG. 6 is a schematic diagram illustrating a module connection of a text structuring apparatus according to a third preferred embodiment of the present invention;
fig. 7 is a schematic diagram illustrating module connections of a text structuring apparatus according to a fourth preferred embodiment of the present invention.
Detailed Description
The following description of the embodiments of the present invention is provided for illustrative purposes, and other advantages and effects of the present invention will become apparent to those skilled in the art from the present disclosure. While the invention will be described in connection with the preferred embodiments, there is no intent to limit its features to those embodiments. On the contrary, the invention is described in connection with the embodiments for the purpose of covering alternatives or modifications that may be extended based on the claims of the present invention. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. The invention may be practiced without these particulars. Moreover, some of the specific details have been left out of the description in order to avoid obscuring or obscuring the focus of the present invention.
Fig. 1 shows a method flowchart of a text structuring method according to a first preferred embodiment of the present invention, which includes step S10, step S20, step S30, and step S40. Specifically, in step S10, the text structuring device 1 obtains the unstructured text, preprocesses the unstructured text, and decomposes the preprocessed unstructured text into a plurality of clauses; in step S20, the text structuring device 1 obtains a question-answer database of structured entries and corresponding structured entries in the structured text; in step S30, the text structuring apparatus 1 matches the content of the clauses to the corresponding structured entries according to the questions in the question-answer database, so as to obtain a clause structured result; in step S40, the text structuring apparatus 1 obtains a structured text from the sentence structuring result.
Here, the text structuring apparatus 1 includes, but is not limited to, a user device, a network device, or a device formed by integrating a user device and a network device through a network. User devices include, but are not limited to, client devices such as computers, smart phones, PDAs, and the like. Network devices include, but are not limited to, computers, network hosts, a single network server, a Cloud of network server sets or servers, where a Cloud is made up of a large number of computers or network servers based on Cloud Computing (Cloud Computing), which is one type of distributed Computing, a virtual supercomputer consisting of a collection of loosely coupled computers. Networks include, but are not limited to, the internet, a wide area network, a metropolitan area network, a local area network, a VPN network, a wireless ad hoc network (ad hoc network), etc.
Of course, those skilled in the art will appreciate that the above-described text structuring apparatus 1 is merely exemplary, and that other existing or future text structuring apparatuses 1, as may be suitable for use in the present application, are also intended to be encompassed within the scope of the present application and are hereby incorporated by reference.
Specifically, in step S10, the text structuring apparatus 1 acquires an unstructured text, pre-processes the unstructured text, and decomposes the pre-processed unstructured text into a plurality of clauses.
The unstructured text is, for example, a descriptive text containing information such as disease symptoms, past medical history, and disease summary in medical records, or a soccer news text or a basketball news text containing information such as the number of goals and the number of attacks of a player, but may be other types of free text.
The text structuring device 1 obtains the unstructured text input by the user, for example, obtains the original medical history text input by the doctor, and performs a preprocessing operation on the unstructured text to convert the irregular unstructured text into the standard unstructured text, thereby facilitating the subsequent conversion operation. Preferably, the pre-treatment comprises: (1) all characters in the unstructured text are converted into full-angle characters, so that the operation is simplified by processing the same type of characters, and the conversion performance is improved; (2) replacing numbers and special symbols in the unstructured text with uniform symbols to simplify subsequent conversion processing, wherein the uniform symbols are symbols which cannot appear in the unstructured text, such as { Number }, { special }, so that numbers and #, such as 10, 1073, 1.763, -0.74 and the like, and special symbols such as @ and the like in the unstructured text are replaced with { Number } and { special }; (3) and removing invisible characters in the unstructured text to simplify the conversion problem and further improve the conversion performance. After the preprocessing, the text structuring device 1 performs sentence-breaking on the preprocessed unstructured text according to the periods in the unstructured text, so as to decompose the unstructured text into a series of clauses. For example, a certain preprocessed unstructured text is "the visible scattered speckle-like density-increasing shadow of the right lung inferior lobe. The remnant lung parenchyma has no definite abnormal density shadow. There was no enlargement of the two lung doors, the trachea and bronchus were unobstructed, and no enlarged lymph nodes were found in the mediastinal septum. There was no fluid accumulation in the pleural cavity on both sides, and no thickening of the pleura. After punctuation marks in the unstructured text are used for sentence breaking, a series of sentences are obtained, namely that the lower lobes of the right lung are visible to be scattered in the speckled density increasing shadow, the parenchyma of the remained lung is not seen with the exact abnormal density shadow, the two lung hilum are not enlarged, the trachea and the bronchus are unobstructed, the mediastinum is not seen with the swollen lymph node, the two side pleural cavities are not seen with the effusion and the pleura is not seen with the thickened.
In step S20, the text structuring apparatus 1 acquires a question-answer database of structured entries and corresponding structured entries in the structured text.
Here, the text structuring apparatus 1 obtains a structured entry in the structured text and a question-and-answer database corresponding to the structured entry, according to the structured text generated as needed. That is to say, the text structuring device 1 first needs to obtain the structured text format that the user wants, and extract each item content in the structured text format from the structured text format, and then set a question according to each item to formulate a question-answer database; or, the structured text format itself acquired by the text structuring device 1 already corresponds to the relevant question-answer data, and then the text structuring device 1 directly acquires each item content in the structured text and directly uses the corresponding question-answer database to perform the subsequent structuring processing.
Preferably, in the preferred embodiment, after the structured entries in the structured text are obtained, the structured entries are classified to obtain classification results by classifying the structured entries into different types, such as values, positions, and the like, and then question templates are set for the classification results, that is, question templates are made for each type of structured entries, and a question-answer database corresponding to the structured entries is formed according to the question templates. Here, the question-answer database can be trained according to a large number of actual texts to continuously add question templates, so that all information in the unstructured texts can be covered by the question templates.
In step S30, the text structuring apparatus 1 matches the content of the clauses to the corresponding structured entries according to the question questions in the question-answer database, respectively, to obtain the clause structured result. For each clause, the question-answer database asks the question, and the content of the clause is matched with the corresponding structured items according to the question result, so that a plurality of clause structured results are obtained.
Specifically, step S30 includes step S31, step S32, and step S33, where in step S31, the text structuring apparatus 1 performs a segmentation process on a sentence, and inputs the obtained segmentation result to the first L STM network, and the first L STM network performs a first decoding process on the segmentation result to obtain a segmentation decoding result, in step S32, the text structuring apparatus 1 generates a question corresponding to the sentence based on the question-and-answer database, and performs a segmentation process on the question, and inputs the obtained question segmentation result to the second L STM network, and the second L STM network performs a second decoding process on the question segmentation result to obtain a question decoding result, and in step S33, the text structuring apparatus 1 performs a combination pairing and modeling of the first L STM network and the second L STM network according to the sentence decoding result and the question decoding result, thereby obtaining the sentence structuring result.
Here, the word segmentation processing is performed on the unstructured text by using a word segmentation algorithm, such as a forward maximum matching word segmentation algorithm, for example, the text structuring apparatus 1 obtains an unstructured medical record text as shown in fig. 2, and performs word segmentation processing on each sentence by using the word segmentation algorithm to obtain a sentence segmentation result as shown in fig. 3, similarly, a question corresponding to each sentence formed based on a question-answer database is also subjected to word segmentation processing by using the word segmentation algorithm to obtain a question word segmentation result, then the sentence segmentation result is input to a first L STM (long-short term memory artificial neural network) network, and the question word segmentation result is input to a second L STM network, wherein the first L STM network performs decoding processing on the sentence segmentation result, and correspondingly, the second L STM network performs decoding processing on the question word segmentation result, and the text structuring apparatus 1 performs modeling processing on the first STM 56 network and the second STM network L STM network based on the segmentation result and the question decoding result, thereby obtaining two STM structures 46L and a combined STM structure.
For each clause, generating a new question according to the current sentence structuring result and the question-answer database, performing further word segmentation processing on the newly generated question, obtaining a new sentence structuring result again, taking the obtained sentence structuring result again as the basis for forming the next question, performing once sentence structuring processing, and repeating the steps until the question and the answer database related to the question and the question are asked completely. That is, for each clause, it is necessary to decide whether to continue to ask the question according to its current structured result, if it is said that the initial structured result of a clause is "there is XX lesion? ", then continue to ask: where is XX the lesion located? How well XX the lesion morphology? Specifically, the initial structuring result as a certain clause is "is whether the lower lobe of the right lung is visible, the focus? "do you continue to ask" see the visible shadow and the focus pattern of the right lung inferior lobe? "etc., until the questions related thereto in the question-and-answer database are asked, to obtain the structured result as shown in fig. 5.
In step S40, the text structuring apparatus 1 obtains a structured text from the sentence structuring result. Preferably, the text structuring device 1 first merges the sentence structuring results to obtain a paragraph structuring result, and then performs post-processing on the paragraph structuring result to obtain a structured text. Here, merging refers to merging all questions with answers in each clause to obtain a final answer, so as to obtain a paragraph structured result, and then processing includes: (1) normalizing the relevant description in the structured text, for example normalizing the I degrees, the one degree, the I degrees and the like in the tonsil size description to be 1 degree, normalizing the 'big bubble sound' in the category of the wet luo sound to be 'rough luo sound', 'middle bubble sound' to be 'middle luo sound', and 'small bubble sound' to be 'thin luo sound'; (2) the numbers and special symbols replaced by the uniform symbols in the preprocessing are subjected to a restoring process, such as restoring 10, 1073, 1.763, -0.74 replaced by the uniform symbols in the previous embodiment to original 10, 1073, 1.763, -0.74, so as to keep the structured text content consistent with the original unstructured text content.
As a modification of the above-described embodiment, as shown in fig. 5, a second preferred embodiment of the present invention provides a text structuring method including step S10 ', step S20', step S30 ', step S40', step S50, and step S60.
Specifically, in step S10', the text structuring device 1 obtains the unstructured text, preprocesses the unstructured text, and decomposes the preprocessed unstructured text into a plurality of clauses; in step S20', the text structuring device 1 obtains a question-answer database of structured entries and corresponding structured entries in the structured text; in step S30', the text structuring apparatus 1 matches the content of the clauses to the corresponding structured entries according to the question questions in the question-and-answer database, respectively, to obtain a clause structured result; in step S40', the text structuring apparatus 1 obtains a structured text according to the sentence structuring result; in step S50, the text structuring device 1 converts the structured text into vectors and stores the vectors in the result database, and performs similarity comparison between the vectors of the structured text and other vectors stored in the result database to obtain similarity text of the structured text; in step S60, the text structuring apparatus 1 calculates the degree of similarity between the structured text and the similar text. Step S10 ', step S20', step S30 'and step S40' are the same as or substantially the same as step S10, step S20, step S30 and step S40 in fig. 1, and thus are not described herein again and are included herein by reference.
In step S50, the text structuring apparatus 1 converts the structured text into vectors and stores the vectors in the result database, and performs similarity comparison between the vectors of the structured text and other vectors stored in the result database to obtain similar text of the structured text.
Here, the euclidean distances between the structured text vector and other text vectors stored in the database are compared, and similarity comparison is performed according to the distance of the euclidean distances to find out similar texts of the structured text from the database. For example, in the medical record, the method can search the medical record text similar to the medical record in the medical record library, thereby facilitating the diagnosis and treatment of the disease for the doctor.
Further, in step S60, the text structuring apparatus 1 calculates the degree of similarity between the structured text and the similar text. That is, for the similar texts retrieved from the database, the text structuring device 1 calculates the similarity between each text and the corresponding structured text, and outputs the similarity to the user, so as to facilitate the comparison and judgment between the texts by the user.
In the medical record text, similarity comparison and similarity calculation can effectively generate similar medical records, and the similarity of the similar medical records is recommended, so that the medical record text can play a great auxiliary role in the work of doctors to better diagnose and treat diseases.
Fig. 6 shows a schematic device diagram of a text structuring device according to a third preferred embodiment of the present invention, where the text structuring device 1 includes a preprocessing module 100, an entry obtaining module 200, a clause structuring module 300, and a text forming module 400. Specifically, the preprocessing module 100 is configured to obtain an unstructured text, preprocess the unstructured text, and decompose the preprocessed unstructured text into multiple clauses; an item obtaining module 200, configured to obtain a structured item in a structured text and a question-answer database corresponding to the structured item; a clause structuring module 300, configured to match the content of the clause to the corresponding structured entries according to the questions in the question-and-answer database, respectively, to obtain a clause structured result; and a text forming module 400, configured to obtain a structured text according to the sentence structuring result.
Here, the text structuring apparatus 1 includes, but is not limited to, a user device, a network device, or a device formed by integrating a user device and a network device through a network. User devices include, but are not limited to, client devices such as computers, smart phones, PDAs, and the like. Network devices include, but are not limited to, computers, network hosts, a single network server, a Cloud of network server sets or servers, where a Cloud is made up of a large number of computers or network servers based on Cloud Computing (Cloud Computing), which is one type of distributed Computing, a virtual supercomputer consisting of a collection of loosely coupled computers. Networks include, but are not limited to, the internet, a wide area network, a metropolitan area network, a local area network, a VPN network, a wireless ad hoc network (ad hoc network), etc.
Of course, those skilled in the art will appreciate that the above-described text structuring apparatus 1 is merely exemplary, and that other existing or future text structuring apparatuses 1, as may be suitable for use in the present application, are also intended to be encompassed within the scope of the present application and are hereby incorporated by reference.
Specifically, the preprocessing module 100 is configured to obtain an unstructured text, preprocess the unstructured text, and decompose the preprocessed unstructured text into multiple clauses.
The unstructured text is, for example, a descriptive text containing information such as disease symptoms, past medical history, and disease summary in medical records, or a soccer news text or a basketball news text containing information such as the number of goals and the number of attacks of a player, but may be other types of free text.
Here, the preprocessing module 100 obtains an unstructured text input by a user, for example, obtains an original medical record text input by a doctor, and performs a preprocessing operation on the unstructured text to convert an irregular unstructured text into a regular unstructured text, thereby facilitating a subsequent conversion operation. Preferably, the pre-treatment comprises: (1) all characters in the unstructured text are converted into full-angle characters, so that the operation is simplified by processing the same type of characters, and the conversion performance is improved; (2) replacing numbers and special symbols in the unstructured text with uniform symbols to simplify subsequent conversion processing, wherein the uniform symbols are symbols which cannot appear in the unstructured text, such as { Number }, { special }, so that numbers and #, such as 10, 1073, 1.763, -0.74 and the like, and special symbols such as @ and the like in the unstructured text are replaced with { Number } and { special }; (3) and removing invisible characters in the unstructured text to simplify the conversion problem and further improve the conversion performance. After the preprocessing, the preprocessing module 100 performs sentence segmentation on the preprocessed unstructured text according to the periods in the unstructured text, so as to decompose the unstructured text into a series of clauses.
For example, a certain preprocessed unstructured text is "the visible scattered speckle-like density-increasing shadow of the right lung inferior lobe. The remnant lung parenchyma has no definite abnormal density shadow. There was no enlargement of the two lung doors, the trachea and bronchus were unobstructed, and no enlarged lymph nodes were found in the mediastinal septum. There was no fluid accumulation in the pleural cavity on both sides, and no thickening of the pleura. After punctuation marks in the non-structural text are used for sentence segmentation, a series of sentences are obtained, wherein the series of sentences are that the lower lobes of the right lung are visible to be scattered in the speckled density increasing shadow, the parenchyma of the remained lung is not seen in the exact abnormal density shadow, the two lung hilum are not enlarged, the trachea and the bronchus are unobstructed, the mediastinum is not seen in the enlarged lymph node, the pleural cavity is not seen in the effusion and the pleura is not seen in the thickening.
An entry obtaining module 200, configured to obtain a question and answer database of structured entries and corresponding structured entries in the structured text.
Here, the entry obtaining module 200 obtains the structured entries in the structured text and the question-answer database corresponding to the structured entries according to the structured text generated as needed, that is, the entry obtaining module 200 first needs to obtain the structured text format desired by the user, extract the content of each entry in the structured text format, and then set questions according to each entry to formulate the question-answer database; or, the structured text format itself acquired by the entry acquiring module 200 already corresponds to the relevant question and answer data, and then the entry acquiring module 200 directly acquires each entry content in the structured text and directly performs the subsequent structuring processing by using the corresponding question and answer database.
Preferably, in this preferred embodiment, the entry obtaining module 200 includes a classifying unit 201 and a forming unit 202, where the classifying unit 201 is configured to, after obtaining the structured entries in the structured text, firstly classify the structured entries to classify the structured entries into different types, such as numerical values, location points, and the like, to obtain classification results, and then the forming unit 202 sets question templates for the classification results respectively, that is, a question template is formulated for each type of structured entries, and a question-and-answer database corresponding to the structured entries is formed according to the question templates. Here, the question-answer database can be trained according to a large number of actual texts to continuously add question templates, so that all information in the unstructured texts can be covered by the question templates.
And a clause structuring module 300, configured to match the content of the clause to the corresponding structured entries according to the questions in the question-answer database, so as to obtain a clause structured result. The question-answer database asks questions of each clause, and the content of each clause is matched with the corresponding structured items according to the question results, so that a plurality of clause structured results are obtained.
Specifically, the sentence structuring module 300 comprises a sentence processing unit 301, a question processing unit 302 and a combined pairing unit 303, wherein the sentence processing unit 301 is configured to perform word segmentation processing on a sentence and input an obtained sentence segmentation result to a first L STM network, the first L STM network performs first decoding processing on the sentence segmentation result to obtain a sentence decoding result, the question processing unit 302 is configured to generate a question corresponding to the sentence based on a question and answer database and perform word segmentation processing on the question, the obtained question segmentation result is input to a second L STM network, the second L STM network performs second decoding processing on the question segmentation result to obtain a question decoding result, and the combined pairing unit 303 is configured to perform combined pairing modeling and modeling on the first L STM network and the second L STM network according to the sentence decoding result and the question decoding result, so that the sentence structuring result is obtained.
Here, the segmentation processing unit 301 performs segmentation processing on an unstructured text using a segmentation algorithm, such as a forward maximum matching segmentation algorithm, for example, to obtain an unstructured medical record text as shown in fig. 2, and performs segmentation processing on each segment using the segmentation algorithm to obtain a segmentation result as shown in fig. 3, similarly, to ask questions corresponding to each segment formed based on a question-answer database, the segmentation processing is also performed using the segmentation algorithm to obtain a question segmentation result, and then, the segmentation processing unit 301 inputs the segmentation result to a first L STM (long-short term memory artificial neural network) network, and the question processing unit 302 inputs the segmentation result of the question to a second L STM network, where the first L STM network performs decoding processing on the segmentation result in the combined matching unit 303, and the second L STM network performs decoding processing on the question segmentation result, and the combined matching unit 303 performs modeling on the first STM network and the STM networks 4656 and 46L based on the segmentation results of the two STM networks, thereby obtaining two STM structures.
For each clause, generating a new question according to the current sentence structuring result and the question-answer database, performing further word segmentation processing on the newly generated question, obtaining a new sentence structuring result again, taking the obtained sentence structuring result again as the basis for forming the next question, performing once sentence structuring processing, and repeating the steps until the question and the answer database related to the question and the question are asked completely. That is, for each clause, it is necessary to decide whether to continue to ask the question according to its current structured result, if it is said that the initial structured result of a clause is "there is XX lesion? ", then continue to ask: where is XX the lesion located? How well XX the lesion morphology? Specifically, the initial structuring result as a certain clause is "is whether the lower lobe of the right lung is visible, the focus? "do you continue to ask" see the visible shadow and the focus pattern of the right lung inferior lobe? "etc., until the questions related thereto in the question-and-answer database are asked, to obtain the structured result as shown in fig. 5.
And a text forming module 400, configured to obtain a structured text according to the sentence structuring result. Preferably, the text forming module 400 includes a merging unit 401 and a post-processing unit 402, wherein the merging unit 401 merges a plurality of sentence structuring results to obtain a paragraph structuring result, and the post-processing unit 402 performs post-processing on the paragraph structuring result to obtain a structured text. Here, merging refers to merging all questions with answers in each clause to obtain a final answer, so as to obtain a paragraph structured result, and then processing includes: (1) normalizing the relevant description in the structured text, for example normalizing the I degrees, the one degree, the I degrees and the like in the tonsil size description to be 1 degree, normalizing the 'big bubble sound' in the category of the wet luo sound to be 'rough luo sound', 'middle bubble sound' to be 'middle luo sound', and 'small bubble sound' to be 'thin luo sound'; (2) the numbers and special symbols replaced by the uniform symbols in the preprocessing are subjected to a restoring process, such as restoring 10, 1073, 1.763, -0.74 replaced by the uniform symbols in the previous embodiment to original 10, 1073, 1.763, -0.74, so as to keep the structured text content consistent with the original unstructured text content.
As a modification of the above embodiment, as shown in fig. 7, a fourth preferred embodiment of the present invention provides a text structuring apparatus, which further includes a similarity judging module 500 and a similarity calculating module 600.
Specifically, the similarity determining module 500 is configured to convert the structured text into a vector, store the vector in the result database, and perform similarity comparison between the vector of the structured text and other vectors stored in the result database to obtain a similarity text of the structured text.
Here, the euclidean distances between the structured text vector and other text vectors stored in the database are compared, and similarity comparison is performed according to the distance of the euclidean distances to find out similar texts of the structured text from the database. For example, in the medical record, the device can search the medical record text similar to the medical record in the medical record library, thereby facilitating the diagnosis and treatment of the disease for the doctor.
Further, the similarity calculating module 600 is configured to calculate a similarity between the structured text and the similar text. That is, for the similar texts retrieved from the database, the text structuring device 1 calculates the similarity between each text and the corresponding structured text, and outputs the similarity to the user, so as to facilitate the comparison and judgment between the texts by the user.
In the medical diagnosis, the similarity comparison and the similarity calculation can effectively generate similar medical records, recommend the similarity of the similar medical records, and play a great auxiliary role in the work of doctors so as to better diagnose and treat diseases.
As a modification of the above embodiment, the present invention also provides a text structuring system including the text structuring device in the above embodiment.
As a modification of the above embodiment, the present invention also provides a nonvolatile storage medium having a text structuring program stored thereon, the text structuring program being executed by a computer to implement a text structuring method, comprising:
the method comprises the steps of a, obtaining an unstructured text, preprocessing the unstructured text, and decomposing the preprocessed unstructured text into a plurality of clauses;
instruction b, obtaining a structured item in the structured text and a question-answer database corresponding to the structured item;
the instruction c is used for respectively matching the content of the clauses to the corresponding structured items according to question questions in the question-answer database so as to obtain a clause structured result;
and d, obtaining a structured text according to the sentence structuring result.
As described above, the text structuring method, apparatus, system and non-volatile storage medium disclosed by the present invention, in combination with the question-answer database, can completely convert unstructured text information into structured information, and have good conversion effect and high accuracy, and perform clause structuring processing through two L STM networks, can process diverse expression modes in free text, and have good robustness.
In summary, the above-mentioned embodiments are provided only for illustrating the principles and effects of the present invention, and not for limiting the present invention. Any person skilled in the art can modify or change the above-mentioned embodiments without departing from the spirit and scope of the present invention. Accordingly, it is intended that all equivalent modifications or changes which can be made by those skilled in the art without departing from the spirit and technical spirit of the present invention be covered by the claims of the present invention.
Claims (12)
1. A method for text structuring, the method comprising:
acquiring an unstructured text, preprocessing the unstructured text, and decomposing the preprocessed unstructured text into a plurality of clauses;
acquiring a structured entry in a structured text and a question-answer database corresponding to the structured entry;
matching the content of the clauses to the corresponding structured entries respectively according to question questions in the question-answer database to obtain clause structured results, including:
performing word segmentation processing on the clauses, inputting the obtained sentence segmentation result into a first L STM network, and performing first decoding processing on the sentence segmentation result by the first L STM network to obtain a sentence decoding result;
generating a question corresponding to the clause based on the question-answer database, performing word segmentation processing on the question, inputting an obtained question word segmentation result to a second L STM network, and performing second decoding processing on the question word segmentation result by the second L STM network to obtain a question decoding result;
and carrying out combined pairing and modeling on the first L STM network and the second L STM network according to the clause decoding result and the problem decoding result so as to obtain a clause structuring result, and obtaining the structured text according to the clause structuring result.
2. The text structuring method according to claim 1, wherein the preprocessing comprises: and replacing the numbers and special symbols in the unstructured text with uniform symbols.
3. The method of claim 1, wherein obtaining a structured entry in the structured text and the question-answer database corresponding to the structured entry comprises:
classifying the structured items to obtain a classification result;
and respectively setting question templates aiming at the classification results, and forming the question-answer database corresponding to the structured entries according to the question templates.
4. The method of claim 1, wherein obtaining the structured text according to the sentence structuring result comprises:
combining a plurality of sentence structuring results to obtain a paragraph structuring result;
and post-processing the paragraph structured result to obtain the structured text.
5. The method of claim 1, wherein after obtaining the structured text according to the sentence structuring result, the method further comprises:
converting the structured text into vectors, storing the vectors in a result database, and performing similarity comparison on the vectors of the structured text and other vectors stored in the result database to obtain similarity texts of the structured text;
and calculating the similarity between the structured text and the similarity text.
6. A text structuring apparatus, comprising:
the preprocessing module is used for acquiring an unstructured text, preprocessing the unstructured text and decomposing the preprocessed unstructured text into a plurality of clauses;
the item acquisition module is used for acquiring a structured item in a structured text and a question-answer database corresponding to the structured item;
a clause structuring module, configured to match the content of the clause to the corresponding structured entries according to the question questions in the question-and-answer database, respectively, so as to obtain a clause structured result, including:
a clause processing unit, configured to perform clause processing on the clause, and input an obtained clause and clause result to a first L STM network, where the first L STM network performs first decoding processing on the clause and clause result to obtain a clause decoding result;
the question processing unit is used for generating questions corresponding to the clauses based on the question-answer database, performing word segmentation on the questions, inputting obtained question word segmentation results to a second L STM network, and performing second decoding processing on the question word segmentation results by the second L STM network to obtain question decoding results;
the system comprises a sentence decoding unit used for decoding a sentence, a sentence combination matching unit used for performing combination matching and modeling on the first L STM network and the second L STM network according to the sentence decoding result and the question decoding result to obtain a sentence structuring result, and a text forming module used for obtaining the structuring text according to the sentence structuring result.
7. The text structuring device as recited in claim 6, wherein the preprocessing module is further configured to replace numbers and special symbols in the unstructured text with uniform symbols.
8. The text structuring device as recited in claim 6, wherein said item obtaining module comprises:
the classification unit is used for classifying the structured items to obtain a classification result;
and the forming unit is used for setting a question template according to the classification result and forming the question-answer database corresponding to the structured entries according to the question template.
9. The text structuring device according to claim 6, wherein the text forming module comprises:
the merging unit is used for merging a plurality of sentence structuring results to obtain paragraph structuring results;
and the post-processing unit is used for post-processing the paragraph structured result to obtain the structured text.
10. The text structuring device according to claim 6, further comprising:
the similarity judging module is used for converting the structured text into vectors and storing the vectors in a result database, and performing similarity comparison on the vectors of the structured text and other vectors stored in the result database to obtain a similarity text of the structured text;
and the similarity calculation module is used for calculating the similarity between the structured text and the similarity text.
11. A text structuring system, characterized in that it comprises a text structuring device according to any one of claims 6-10.
12. A non-volatile storage medium having a text structuring program stored thereon, the text structuring program being executed by a computer to implement a text structuring method, comprising:
the method comprises the steps of a, obtaining an unstructured text, preprocessing the unstructured text, and decomposing the preprocessed unstructured text into a plurality of clauses;
the instruction b is used for acquiring a structured entry in the structured text and a question-answer database corresponding to the structured entry;
and c, respectively matching the content of the clauses to the corresponding structured entries according to the question questions in the question-answer database to obtain a clause structured result, wherein the instruction c comprises the following steps:
performing word segmentation processing on the clauses, inputting the obtained sentence segmentation result into a first L STM network, and performing first decoding processing on the sentence segmentation result by the first L STM network to obtain a sentence decoding result;
generating a question corresponding to the clause based on the question-answer database, performing word segmentation processing on the question, inputting an obtained question word segmentation result to a second L STM network, and performing second decoding processing on the question word segmentation result by the second L STM network to obtain a question decoding result;
and performing combined pairing and modeling on the first L STM network and the second L STM network according to the clause decoding result and the problem decoding result to obtain a clause structuring result, and obtaining the structured text according to the clause structuring result by an instruction d.
Priority Applications (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010477184.6A CN111680089B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
| CN202010477190.1A CN111680090B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
| CN202010511844.8A CN111680094B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
| CN201710852183.3A CN107729392B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710852183.3A CN107729392B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
Related Child Applications (3)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010511844.8A Division CN111680094B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
| CN202010477184.6A Division CN111680089B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
| CN202010477190.1A Division CN111680090B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107729392A CN107729392A (en) | 2018-02-23 |
| CN107729392B true CN107729392B (en) | 2020-07-10 |
Family
ID=61206611
Family Applications (4)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010477190.1A Active CN111680090B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
| CN202010477184.6A Active CN111680089B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
| CN202010511844.8A Active CN111680094B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
| CN201710852183.3A Active CN107729392B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
Family Applications Before (3)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010477190.1A Active CN111680090B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
| CN202010477184.6A Active CN111680089B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
| CN202010511844.8A Active CN111680094B (en) | 2017-09-19 | 2017-09-19 | Text structuring method, device and system and non-volatile storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (4) | CN111680090B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108711443B (en) * | 2018-05-07 | 2021-11-30 | 成都智信电子技术有限公司 | Text data analysis method and device for electronic medical record |
| CN108629019B (en) * | 2018-05-08 | 2021-04-30 | 桂林电子科技大学 | Question-answer field-oriented question sentence similarity calculation method containing names |
| CN110472925A (en) * | 2018-05-11 | 2019-11-19 | 懿谷智能科技(上海)有限公司 | A kind of laboratory test process management system and method based on webpage flow chart |
| CN108733837B (en) * | 2018-05-28 | 2021-04-27 | 上海依智医疗技术有限公司 | Natural language structuring method and device for medical history text |
| CN109145299B (en) * | 2018-08-16 | 2022-06-21 | 北京金山安全软件有限公司 | Text similarity determination method, device, equipment and storage medium |
| CN109409645A (en) * | 2018-09-07 | 2019-03-01 | 平安科技（深圳）有限公司 | The method and storage medium that electronic device, lawyer recommend |
| CN109493926A (en) * | 2018-10-30 | 2019-03-19 | 中山大学肿瘤防治中心 | Processing method, device, medium and the electronic equipment of colorectal cancer medical data |
| CN109800284B (en) * | 2018-12-19 | 2021-02-05 | 中国电子科技集团公司第二十八研究所 | Task-oriented unstructured information intelligent question-answering system construction method |
| CN110415791A (en) * | 2019-01-29 | 2019-11-05 | 四川大学华西医院 | System and method is established in a kind of disease library |
| CN110321466B (en) * | 2019-06-14 | 2023-09-15 | 广发证券股份有限公司 | Securities information duplicate checking method and system based on semantic analysis |
| CN110909137A (en) * | 2019-10-12 | 2020-03-24 | 平安科技（深圳）有限公司 | Information pushing method and device based on man-machine interaction and computer equipment |
| CN111125100A (en) * | 2019-12-12 | 2020-05-08 | 东软集团股份有限公司 | Data storage method and device, storage medium and electronic equipment |
| CN112765194B (en) * | 2020-12-31 | 2024-04-30 | 科大讯飞股份有限公司 | Data retrieval method and electronic equipment |
| CN112364035A (en) * | 2021-01-14 | 2021-02-12 | 零犀(北京)科技有限公司 | Processing method and device for call record big data, electronic equipment and storage medium |
| CN112800759B (en) * | 2021-04-14 | 2021-08-06 | 北京金山云网络技术有限公司 | Standardized data generation method and device and medical text data processing method and device |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1497473A (en) * | 2002-09-30 | 2004-05-19 | Metod and device for text structurng | |
| CN104899260A (en) * | 2015-05-20 | 2015-09-09 | 东华大学 | Method for structured processing of Chinese pathological text |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10409908B2 (en) * | 2014-12-19 | 2019-09-10 | Google Llc | Generating parse trees of text segments using neural networks |
| CN106095913A (en) * | 2016-06-08 | 2016-11-09 | 广州同构医疗科技有限公司 | A kind of electronic health record text structure method |
| CN106649561B (en) * | 2016-11-10 | 2020-05-26 | 复旦大学 | Intelligent question-answering system for tax consultation service |
| CN106897568A (en) * | 2017-02-28 | 2017-06-27 | 北京大数医达科技有限公司 | The treating method and apparatus of case history structuring |
-
2017
- 2017-09-19 CN CN202010477190.1A patent/CN111680090B/en active Active
- 2017-09-19 CN CN202010477184.6A patent/CN111680089B/en active Active
- 2017-09-19 CN CN202010511844.8A patent/CN111680094B/en active Active
- 2017-09-19 CN CN201710852183.3A patent/CN107729392B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1497473A (en) * | 2002-09-30 | 2004-05-19 | Metod and device for text structurng | |
| CN104899260A (en) * | 2015-05-20 | 2015-09-09 | 东华大学 | Method for structured processing of Chinese pathological text |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111680090B (en) | 2023-03-21 |
| CN111680089A (en) | 2020-09-18 |
| CN111680094B (en) | 2023-03-21 |
| CN107729392A (en) | 2018-02-23 |
| CN111680089B (en) | 2023-03-21 |
| CN111680094A (en) | 2020-09-18 |
| CN111680090A (en) | 2020-09-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107729392B (en) | Text structuring method, device and system and non-volatile storage medium | |
| US20200184276A1 (en) | Method and system for generating and correcting classification models | |
| US11210468B2 (en) | System and method for comparing plurality of documents | |
| CN108681557B (en) | Short text topic discovery method and system based on self-expansion representation and similar bidirectional constraint | |
| US9183285B1 (en) | Data clustering system and methods | |
| Franzoni et al. | Context-based image semantic similarity | |
| SzymańSki | Comparative analysis of text representation methods using classification | |
| CN111768869B (en) | Medical guide mapping construction search system and method for intelligent question-answering system | |
| US11170169B2 (en) | System and method for language-independent contextual embedding | |
| Altheneyan et al. | Big data ML-based fake news detection using distributed learning | |
| CN113688624A (en) | Personality prediction method and device based on language style | |
| Ahmed et al. | Developing an ontology of concepts in the Qur'an | |
| Contreras et al. | Using topic modelling for analyzing panamanian parliamentary proceedings with neural and statistical methods | |
| Kumar et al. | Efficient structuring of data in big data | |
| Cardenas et al. | Improving Topic Coherence Using Entity Extraction Denoising. | |
| Akhgari et al. | Sem-TED: semantic twitter event detection and adapting with news stories | |
| Bembenik et al. | Intelligent methods and big data in industrial applications | |
| Wang et al. | A semantic path based approach to match subgraphs from large financial knowledge graph | |
| CN105808522A (en) | Method and apparatus for semantic association | |
| Zouaoui et al. | Ontological Approach Based on Multi-Agent System for Indexing and Filtering Arabic Docu-ments | |
| Duque et al. | A multiview clustering approach for mining authorial affinities in literary texts | |
| Yuan et al. | Self-adaptive extracting academic entities from World Wide Web | |
| Makruf et al. | Public hospital review on map service with part of speech tagging and biterm topic modeling | |
| Ning | Research on the extraction of accounting multi-relationship information based on cloud computing and multimedia | |
| Khan et al. | Sentiment Analysis using Support Vector Machine and Random Forest |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |