A kind of checking method and system of forensic identification reportTechnical field
The present invention relates to extraction of semantics and analysis field, the especially a kind of checking method and system of forensic identification report.
Background technique
Under the background of current reform of the judicial system iterative method, a large amount of survey reports, expert opinion, assessment report and speciallyProfessional, scientific and objectivity the requirement of family's examination reports is higher and higher, and procurator, lawyer and judge are to suchThe examination consciousness and depth of evidence are constantly reinforced.But such opinion or report still rely on investigation appraiser with this profession artLanguage is write using all kinds of text edit softwares, exists and searches and manage bring time-consuming by artificial because of cited data sourceLong, the problem of process is many and diverse, low efficiency.For example, the figure in the law circle that Yunnan Province handles drug-related case is mostly to be related to the life of drugsObject chemistry expert can identify case sample of being involved in drug traffic, and write probation report, but interior ground province method using softwares such as WordRule personage does not have related drugs identification technical ability, so needing to rely on social appraisal organization provides probation report.Law attainment, prisonSuperintending and directing many reasons such as management causes the conclusion of social appraisal organization there is certain flaw and mistake, for example identification item is not completeThe relevance in face, identification and case is not strong etc.；Meanwhile there are following characteristics for forensic identification report: term is that clause is special in contentSign, Similar content is more, and length is longer, causes semantic analysis accuracy rate low；Therefore a kind of checking method and system, accurate language are neededJustice is analyzed, the mistake in examination report, completes the report identification of high accuracy.
Summary of the invention
It is an object of the invention to: the present invention provides the checking methods and system of a kind of report of forensic identification, help methodRule personage quickly and accurately examines flaw present in probation report and mistake.
The technical solution adopted by the invention is as follows:
A kind of checking method of forensic identification report, includes the following steps:
Step 1: after the multidimensional characteristic using case legal medical expert's probation report in NLP extraction database, using word-based frequency divisionThe HMM model that cloth and the multistage Markov of synonym analysis are assumed carries out multidimensional characteristic segmentation and fitting, obtains and stores spyLevy model；
Step 2: after inputting pending report, assignment options being examined based on interactive unit selection, characteristic model is called to carry out latticeFormula examines and damage location examines；
Step 3: the examination result examined according to formality examination and damage location generates and shows legal medical expert's Identification and investigation reportIt accuses.
Preferably, the step 1 includes the following steps:
Step 1.1: the multidimensional characteristic of body matter in case legal medical expert probation report, the multidimensional characteristic are extracted using NLPIncluding case description, medicolegal examination, analytic explanation and expert's conclusion；
Step 1.2: case property, case plot, case time of origin and the case duration data in case description are extracted,The HMM model assumed using the multistage Markov based on word frequency distribution is formed to that can be split with term in case descriptionSerialize feature, that is, word frequency distribution and relevance data；
Step 1.3: the term in medicolegal examination is extracted, using based on the more of the word frequency distribution after synonym analysis correctionsThe HMM model that rank Markov is assumed, is identified and is marked to the term in medicolegal examination；Using OCR technique to legal medical expertThe image data learned in examining is identified, is marked, and checks image data format, and analytical analysis illustrates to obtain a plurality of termTable, composition sequence feature；
Step 1.4: after expert's conclusion is carried out token segmentation, the high template library of matching similarity, and it is stored in the mouldPlate library；
Step 1.5: the resulting serializing feature of fit procedure 1.2-1.4 and data composition characteristic model.
Preferably, formality examination includes the following steps: in the step 2
Step a1: acquiring the body matter and attachment content of pending report, and body matter includes title, introduction, medical jurisprudenceInspection, analytic explanation, expert's conclusion, surveyor and appraisal organization；Attachment content includes physical examination, presentation content, operation noteRecord, medicolegal examination and medical jurisprudence image data；
Step a2: body matter and attachment content are serialized, and memory headroom is stored in the form of characteristic, are calledCharacteristic model is compared and analyzes.
Preferably, damage location examination includes the following steps: in the step 2
Step b1: acquiring the attachment content in pending report, and the attachment content includes physical examination, presentation content, handArt record, medicolegal examination, medical jurisprudence image data；
Step b2: characteristic model is called, damage location is examined using knowledge mapping therein.
A kind of auditing system of forensic identification report, including
Database, for storing case legal medical expert probation report, characteristic model and knowledge mapping；
Model foundation unit makes for extracting the corresponding multidimensional characteristic of case legal medical expert's probation report in database using NLPMultidimensional characteristic segmentation and fitting are carried out with the HMM model that the multistage Markov based on word frequency distribution and synonym analysis is assumed,Construction feature model；
Interactive unit examines assignment options for inputting pending report and selection, and shows that legal medical expert's Identification and investigation is reported；
It examines unit, after carrying out data processing to pending report, characteristic model is called to carry out formality examination and damagePosition examines.
Preferably, the multidimensional characteristic in body matter that the model foundation unit extracts includes case description, medical jurisprudenceInspection, analytic explanation and expert's conclusion.
Preferably, the model foundation unit includes
Case Expressive Features extraction module, the HMM model for using the multistage Markov based on word frequency distribution to assume,To that can be split with term in case description, forms word frequency distribution and relevance data serializes feature；
Medicolegal examination characteristic extracting module, for using the multistage horse based on the word frequency distribution after synonym analysis correctionsThe HMM model that Er Kefu assumes, is identified and is marked to the term in medicolegal examination；Medical jurisprudence is examined using OCR techniqueImage data in testing is identified, is marked, and checks image data format, and analytical analysis illustrates to obtain a plurality of nomenclature, groupAt serializing feature；
Expert's conclusion characteristic extracting module, after carrying out token segmentation, the high template library of matching similarity, and storeIn the template library；
Feature fitting module, the feature extracted for being fitted all characteristic extracting modules；
The case Expressive Features extraction module, medicolegal examination characteristic extracting module, expert's conclusion characteristic extracting moduleIt is connect respectively with feature fitting module.
Preferably, the examination unit includes that formality examination module and damage location examine module.
Preferably, the formality examination module include the pending report for that will acquire body matter and attachment content intoThe data processing module of row serializing.
Preferably, the damage location examines that module includes the read module for examining attachment content in pending report.
In conclusion by adopting the above-described technical solution, the beneficial effects of the present invention are:
1. the present invention is examined by characteristic model unified standard, auxiliary personnel in charge of the case carries out electronic authentication, reduces reportIt accuses because of identification side's bring mistake and error, injury of human degree Identification and investigation is standardized, modularization, personnel in charge of the case canIndependent operation completes the judgement to professional accreditation conclusion；
2. assisted review experience is incorporated characteristic model assisted review by the present invention, it is conducive to newly fast into legal medical expert personnel and ordinary peopleSpeed grasps checking method, improves and examines efficiency；
3. the present invention constructs generic term synonym and context-sensitive characteristic using in medical jurisprudence by buildingMultistage Markov assume HMM model, using based on word frequency distribution multistage Markov assume HMM model to caseIt can be split with term in description, be assumed using the multistage Markov based on the word frequency distribution after synonym analysis correctionsHMM model is identified and is marked to the term in medicolegal examination, and feature extraction precision is greatly improved, to improve legal medical expert's mirrorSurely the accuracy of examination reports is reported.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below will be to needed in the embodiment attachedFigure is briefly described, it should be understood that the following drawings illustrates only certain embodiments of the present invention, therefore is not construed as pairThe restriction of range for those of ordinary skill in the art without creative efforts, can also be according to thisA little attached drawings obtain other relevant attached drawings.
Fig. 1 is flow chart of the method for the present invention；
Fig. 2 is the block diagram of system of the invention；
Fig. 3 is characteristic model schematic diagram of the invention；
Fig. 4 is examination report schematic diagram of the invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, rightThe present invention is further elaborated.It should be appreciated that described herein, specific examples are only used to explain the present invention, notFor limiting the present invention, i.e., described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is logicalThe component for the embodiment of the present invention being often described and illustrated herein in the accompanying drawings can be arranged and be designed with a variety of different configurations.
Therefore, the detailed description of the embodiment of the present invention provided in the accompanying drawings is not intended to limit below claimedThe scope of the present invention, but be merely representative of selected embodiment of the invention.Based on the embodiment of the present invention, those skilled in the artMember's every other embodiment obtained without making creative work, shall fall within the protection scope of the present invention.
It should be noted that the relational terms of term " first " and " second " or the like be used merely to an entity orOperation is distinguished with another entity or operation, and without necessarily requiring or implying between these entities or operation, there are anyThis actual relationship or sequence.Moreover, the terms "include", "comprise" or its any other variant be intended to it is non-exclusiveProperty include so that include a series of elements process, method, article or equipment not only include those elements, but alsoFurther include other elements that are not explicitly listed, or further include for this process, method, article or equipment it is intrinsicElement.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that including describedThere is also other identical elements in the process, method, article or equipment of element.
Feature and performance of the invention are described in further detail with reference to embodiments.
Embodiment 1
There is certain flaw and mistakes for the conclusion of existing society's appraisal organization, for example identification item is not comprehensive, identificationIt is not strong etc. with the relevance of case；Meanwhile there are following characteristics for forensic identification report: term is sentence features in content, similarContent is more, and length is longer, causes semantic analysis accuracy rate low；Therefore the application proposes a kind of checking method and system, accurate languageJustice is analyzed, the mistake in examination report, completes the report identification of high accuracy, and details is as follows:
As shown in Figure 1, a kind of checking method of forensic identification report, characterized by the following steps:
Step 1: the multidimensional characteristic of the case information stored in database is extracted using NLP, fitting multidimensional characteristic obtains specialLevy model；
Step 2: after inputting pending report, assignment options being examined based on interactive unit selection, call the feature in databaseModel carries out formality examination and damage location examines；
Step 3: obtaining examination result, generate and show that legal medical expert's Identification and investigation is reported.
Because there are following characteristics for forensic identification report: similar item is more, cause Similar content can not duplicate removal, to the heavier condition of the injury,The part such as length longer expert opinion can judge to be not allowed when extracting to condition of the injury position, cause to judge by accident；Term is that clause is specialIt levies, there are relevances between the extraction process of sample, the number of sample are read and identified the contextual feature for having all kinds of samples.CauseThis application building utilizes the multistage Ma Erke constructed in medical jurisprudence to generic term synonym and context-sensitive characteristicHusband assume HMM model, using based on word frequency distribution multistage Markov assume HMM model to case description in can use artLanguage is split, and the HMM model assumed using the multistage Markov based on the word frequency distribution after synonym analysis corrections is to methodTerm in medical test is identified and is marked, and feature extraction precision is greatly improved, and is examined to improve forensic identification reportThe accuracy of opinion.
Wherein, hidden Markov model (Hidden Markov Model, HMM) is the probabilistic model about timing, descriptionIt generates unobservable state random sequence at random by a hiding Markov chain, then an observation is generated by each stateAnd generate the process of observation random sequence.The sequence for the state that hiding Markov chain generates at random, referred to as status switch(state sequence)；Each state generates one and observes, and the random sequence of resulting observation, referred to as observation sequence(observation sequence).Each position of sequence can be regarded as a moment again.
When construction feature model, the multidimensional characteristic of body matter in case legal medical expert probation report is extracted using NLP, it is described moreDimensional feature includes case description, medicolegal examination, analytic explanation and expert's conclusion；
Using based on word frequency distribution multistage Markov assume HMM model, to case description in can be carried out with termSegmentation, word frequency distribution and relevance data serialize feature.
The HMM model assumed using the multistage Markov based on the word frequency distribution after synonym analysis corrections, to legal medical expertThe term learned in examining is identified and is marked.
Image data is identified using OCR technique, marks and check image data format, content is with the presence or absence of mistakeAccidentally, the content that will identify that is compared with the term that previous step identifies, is as a result stored with array form.For existingInconsistent situation is prompted.
To expert's conclusion, the keyword of standard qualification result template is used to carry out semantization segmentation as token, according to woundFeelings position is divided into a plurality of.Structural data, that is, characteristic model after the above serializing.Using OCR technique to the text in attachment contentWord identification, together with body matter word segment using token segmentation after, carry out synonym serializing, by obtained data withThe form of Multidimensional numerical is stored in memory, when a plurality of preliminary conclusion is fitted, needs to utilize point based on vector spaceAnalyse model.
Formality examination includes the following steps:
Step a1: acquiring the body matter and attachment content of pending report, and body matter includes title, introduction, medical jurisprudenceInspection, analytic explanation, expert's conclusion, surveyor and appraisal organization；Attachment content includes physical examination, presentation content, operation noteRecord, medicolegal examination and medical jurisprudence image data；
Step a2: body matter and attachment content are serialized, and memory headroom is stored in the form of characteristic, are calledCharacteristic model is compared and analyzes.
Damage location examination includes the following steps:
Step b1: acquiring the attachment content in pending report, and the attachment content includes physical examination, presentation content, handArt record, medicolegal examination, medical jurisprudence image data；
Step b2: characteristic model is called, damage location is examined using knowledge mapping therein.Wherein, according to " peopleBulk damage standard of perfection " classification of partes corporis humani's division methods, knowledge mapping is generated using 190 remainder of totally 12 class damage item as node.
As shown in figure 3, imaging data copy be shown as " rib cage multiple fracture " then sequence turn to keyword " rib cage "," at two or more than fracture " are matched to the characteristic model of regular 5.6.4b, so analyze be " slight wound second level " preliminary conclusion.RootThe examination report obtained according to the application is as shown in figure 4, include formality examination opinion and damage location opinion, the application reflects to legal medical expertFixed technical examination result improves 17% or so accuracy.
Embodiment 2
Based on the method for embodiment 1, the present embodiment provides a kind of corresponding system, details is as follows:
As shown in Fig. 2, a kind of auditing system of forensic identification report, including
Database, for storing case legal medical expert probation report, characteristic model and knowledge mapping；
Model foundation unit makes for extracting the corresponding multidimensional characteristic of case legal medical expert's probation report in database using NLPMultidimensional characteristic segmentation and fitting are carried out with the HMM model that the multistage Markov based on word frequency distribution and synonym analysis is assumed,Construction feature model；
Interactive unit examines assignment options for inputting pending report and selection, and shows that legal medical expert's Identification and investigation is reported；
It examines unit, after carrying out data processing to pending report, characteristic model is called to carry out formality examination and damagePosition examines.
The multidimensional characteristic in body matter that the model foundation unit extracts includes case description, medicolegal examination, dividesAnalysis explanation and expert's conclusion.
The model foundation unit includes
Case Expressive Features extraction module, the HMM model for using the multistage Markov based on word frequency distribution to assume,To that can be split with term in case description, forms word frequency distribution and relevance data serializes feature；
Medicolegal examination characteristic extracting module, for using the multistage horse based on the word frequency distribution after synonym analysis correctionsThe HMM model that Er Kefu assumes, is identified and is marked to the term in medicolegal examination；Medical jurisprudence is examined using OCR techniqueImage data in testing is identified, is marked, and checks image data format, and analytical analysis illustrates to obtain a plurality of nomenclature, groupAt serializing feature；
Expert's conclusion characteristic extracting module, after carrying out token segmentation, the high template library of matching similarity, and storeIn the template library；
Feature fitting module, the feature extracted for being fitted all characteristic extracting modules；
The case Expressive Features extraction module, medicolegal examination characteristic extracting module, expert's conclusion characteristic extracting moduleIt is connect respectively with feature fitting module.
The examination unit includes that formality examination module and damage location examine module.
The formality examination module includes that the body matter of the pending report for that will acquire and attachment content carry out sequenceThe data processing module of change.
The damage location examines that module includes the read module for examining attachment content in pending report.
The system comprises processor, memory and store the calculating that can be run in the memory and on a processorMachine program, such as " step 1: the multidimensional characteristic of the case information stored in database is extracted using NLP, fitting multidimensional characteristic obtainsTake characteristic model；Step 2: after inputting pending report, assignment options being examined based on interactive unit selection, call the spy in databaseIt levies model and carries out formality examination and damage location examination；Step 3: obtaining examination result, generate and show legal medical expert's Identification and investigation reportAnnouncement " program, computer program can be divided into one or more module/units, one or more of modules/unit quiltStorage in the memory, and is executed by the processor, to complete the present invention.One or more of module/units canTo be the series of computation machine program instruction section that can complete specific function, which exists for describing the computer programImplementation procedure in system.For example, the computer program can be divided into model foundation unit, interactive unit, examine listMember and database, database, for storing case legal medical expert probation report, characteristic model and knowledge mapping；Model foundation unit is usedIn extracting the corresponding multidimensional characteristic of case legal medical expert's probation report in database using NLP, using based on word frequency distribution and synonymThe HMM model that the multistage Markov of analysis is assumed carries out multidimensional characteristic segmentation and fitting, construction feature model；Interactive unit,Assignment options are examined for inputting pending report and selection, and show that legal medical expert's Identification and investigation is reported；Unit is examined, for pendingAfter report carries out data processing, characteristic model is called to carry out formality examination and damage location examination.The model foundation unit mentionsThe multidimensional characteristic in body matter taken includes case description, medicolegal examination, analytic explanation and expert's conclusion.Model foundation listMember includes case Expressive Features extraction module, and the HMM model for using the multistage Markov based on word frequency distribution to assume is rightIt can be split with term in case description, form word frequency distribution and relevance data serializes feature；Medicolegal examination featureExtraction module, the HMM model for using the multistage Markov based on the word frequency distribution after synonym analysis corrections to assume are rightTerm in medicolegal examination is identified and is marked；The image data in medicolegal examination is identified using OCR technique,Label, and check image data format, analytical analysis illustrates to obtain a plurality of nomenclature, composition sequence feature；Expert's conclusion is specialExtraction module is levied, after carrying out token segmentation, the high template library of matching similarity, and it is stored in the template library；Feature is quasi-Block is molded, the feature extracted for being fitted all characteristic extracting modules；The case Expressive Features extraction module, medicolegal examinationCharacteristic extracting module, expert's conclusion characteristic extracting module are connect with feature fitting module respectively.The examination unit includes formatExamine that module and damage location examine module.The system can be desktop PC, notebook, palm PC and cloud clothesBusiness device etc. calculates equipment.The system may include, but be not limited only to, processor, memory.Those skilled in the art can manageSolution, the schematic diagram is only the example of the system, does not constitute the restriction to the system, may include more than illustratingOr less component, certain components or different components are perhaps combined, such as the system equipment can also include that input is defeatedEquipment, network access equipment, bus etc. out.
Alleged processor can be central processing unit (Central Processing Unit, CPU), can also be itHis general processor, digital signal processor (Digital Signal Processor, DSP), specific integrated circuit(Application Specific Integrated Circuit, ASIC), ready-made programmable gate array (Field-Programmable Gate Array, FPGA) either other programmable logic device, discrete gate or transistor logic,Discrete hardware components etc..General processor can be microprocessor or the processor is also possible to any conventional processorDeng the processor is the control centre of the increase resolution system, utilizes various interfaces and the entire resolution ratio of connectionThe various pieces of lifting system.
The memory can be used for storing the computer program and/or module, and the processor is by operation or executesComputer program in the memory and/or module are stored, and calls the data being stored in memory, described in realizationThe various functions of increase resolution system.The memory can mainly include storing program area and storage data area, wherein storageIt program area can application program needed for storage program area, at least one function (such as sound-playing function, image player functionDeng) etc.；Storage data area, which can be stored, uses created data (such as audio data, phone directory etc.) etc. according to mobile phone.ThisOutside, memory may include high-speed random access memory, can also include nonvolatile memory, such as hard disk, memory, insertConnect formula hard disk, intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash memoryBlock (Flash Card), at least one disk memory, flush memory device or other volatile solid-state parts.
The application building is multistage using constructing in medical jurisprudence to generic term synonym and context-sensitive characteristicThe HMM model that Markov is assumed, the HMM model assumed using the multistage Markov based on word frequency distribution is in case descriptionIt can be split with term, the HMM mould assumed using the multistage Markov based on the word frequency distribution after synonym analysis correctionsType is identified and is marked to the term in medicolegal examination, and feature extraction precision is greatly improved, to improve forensic identification reportAssisted review experience is incorporated characteristic model assisted review by the accuracy for accusing examination reports, is conducive to newly into legal medical expert personnel and commonPeople quickly grasps checking method, improves and examines efficiency.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the inventionMade any modifications, equivalent replacements, and improvements etc., should all be included in the protection scope of the present invention within mind and principle.