CN102147731A - Automatic functional requirement extraction system based on extended functional requirement description framework - Google Patents

Automatic functional requirement extraction system based on extended functional requirement description framework Download PDF

Info

Publication number
CN102147731A
CN102147731A CN 201110099316 CN201110099316A CN102147731A CN 102147731 A CN102147731 A CN 102147731A CN 201110099316 CN201110099316 CN 201110099316 CN 201110099316 A CN201110099316 A CN 201110099316A CN 102147731 A CN102147731 A CN 102147731A
Authority
CN
Grant status
Application
Patent type
Prior art keywords
functional requirement
functional requirements
module
efrf
grammatical
Prior art date
Application number
CN 201110099316
Other languages
Chinese (zh)
Inventor
唐琦
王楷翔
王英林
郭俊
郭健美
Original Assignee
上海交通大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date

Links

Abstract

The invention discloses an automatic functional requirement extraction system based on an extended functional requirement description framework, belonging to the technical field of computer application. The automatic functional requirement extraction system comprises a grammatical analysis module and a functional requirement conversion module, wherein the grammatical analysis module is based on a grammatical analyzer; the functional requirement conversion module is used for mapping an analysis result of the grammatical analyzer to an EFRF (Extended Functional Requirements Framework) according to a predefined conversion rule, and the analysis result is analyzed and converted by the grammatical analysis module and the functional requirement conversion module. According to the automatic functional requirement extraction system, a functional requirement file is input and analyzed on the basis of the grammatical analyzer; the grammatical analysis result is converted by using a series predefined conversion rules and is mapped to the EFRF; and software functional requirement structural expression is output in an XML (Extensible Markup Language) form and is stored in the system.

Description

基于扩展功能需求描述框架的功能需求自动抽取系统 Functional requirements based on the extended functional requirements of the framework described automatic extraction system

技术领域 FIELD

[0001] 本发明涉及一种计算机应用技术领域的装置,具体是基于扩展的功能需求描述框架(Extended Functional Requirements Framework, EFRF)的功能需求自动抽取系统。 [0001] The present invention relates to a computer technical field application means, in particular based on the described frame expanded functional requirements (Extended Functional Requirements Framework, EFRF) automatic extraction system functional requirements.

背景技术 Background technique

[0002] 软件功能需求描述了一个系统的外部功能。 [0002] software functional requirements describes the external features of a system. 现有的软件功能需求大多以文本形式描述在文档中。 Most existing software functional requirements described in text form in the document. 由于缺乏结构化表达,这些功能需求文档很难有效利用到后续软件开发过程中。 Due to the lack of structural expression, these functional requirements document is difficult to follow the effective use of the software development process. 因此,如何从这些功能需求文档中自动抽取结构化的功能需求表达,有助于提高功能需求的利用率,提高软件开发的效率。 Therefore, how to automatically extract structured functional requirements expressed from these functional requirements document to help improve the utilization of functional requirements, improve the efficiency of software development.

[0003] 经过对现有技术的检索发现,N. Niu和S. Easterbrook人的“抽取和建模产品线功能需求,,("Extracting and modeling product line functional requirements,,,发表于2008年在西班牙巴塞罗那召开的第16届需求工程国际会议集,Proceedings of 2008 International Requirements Engineering Conference,第155-164 页)公开了一禾中半自动化的软件产品线功能需求抽取和建模方法。 [0003] After retrieval of prior art discovery, N. Niu and S. Easterbrook's "extraction and modeling product line functional requirements ,, (" Extracting and modeling product line functional requirements ,,, published in 2008 in Spain set in the 16th international Conference on requirements Engineering held in Barcelona, ​​Proceedings of 2008 international requirements Engineering Conference, pp. 155-164) discloses a half Wo automated software product line feature extraction and modeling needs. 该方法在动宾关系对的基础上,提出了一个包括6个描述维度:施动者(agentive),对象(objective),地点(Iocational),时间(temporal),处理过程(process)和条件(conditional)功能需求可变性模型。 The method based on verb-object relationship between the proposed includes six Description dimensions: doer (agentive), the object (Objective), location (Iocational), time (temporal), process (process), and conditions ( conditional) functional requirements variability model. 缺点是:其可变性模型的定义中还存在不清晰的地方。 The disadvantages are: define its variability of the model there is not a clear place. 例如,“Agentive”和“Objective”的修饰语没有列入维度;“ft^cess”太具有一般性,包含信息不够细节化;在抽取verb-directObject 关系对的时候,采用统计方法,结果不够准确。 For example, "Agentive" and "Objective" qualifier is not included in the dimensions; "ft ^ cess" has too general, contains details of the information is not enough; in relation to the extraction verb-directObject time, the use of statistical methods, the result is not accurate enough .

[0004] Liaskos等人的“基于目标的可变性获取和分析”(“On goal-based variability acquisition and analysis”,发表于2006年在美国明尼阿波利斯召开的第14届需求工禾呈国际会议集,Proceedings of 2006 International Requirements Engineering Conference,第76-85页)定义了一个具有可变性的目标模型,该模型基于Fillmore提出的Case Grammar Theory。 [0004] Liaskos et al, "based on the target variability acquisition and analysis" ( "On goal-based variability acquisition and analysis", published in 2006, held in Minneapolis, the 14th it was international demand for workers meeting set, Proceedings of 2006 International Requirements Engineering Conference, pp. 76-85) defines a target variability model, based on the model proposed by Fillmore Case Grammar Theory. 他们关注于每个目标的OR-组合的语义特征,并且用三个语义维度定义了目标的上下文的可变性。 They focus on semantic features OR- combination of each target, and defines the variability context semantic object with three dimensions. 文章中还强调了这三个语义维度对正交可变性模型(Orthogonal variability models)建模的重要性。 The article also stressed the importance of these three semantic dimensions of the orthogonal variability model (Orthogonal variability models) modeling. 将目标的语义描述维度作为可变点中的变量,在此基础上进行正交可变性建模能更好地体现可变性及其如何可变。 The semantic description of the dimensions of the target point as a variable in the variable orthogonal variability modeling on the basis of better reflect variability and how variable.

[0005] 上述提出的方法还不能实现对软件功能需求的全面的、自动的抽取。 Methods [0005] proposed above can not achieve a comprehensive, demand for automatic extraction software features. 因此,目前软件功能需求的构建仍需要付出大量的人工劳动。 Therefore, the current demand for software functional building still need to pay a lot of manual labor.

发明内容 SUMMARY

[0006] 本发明针对现有技术存在的上述不足,提供一种基于扩展功能需求描述框架的功能需求自动抽取系统,将功能需求文档作为输入后基于语法分析器对功能需求文档进行分析,通过一系列预定义的转换规则对语法分析结果进行转换,映射到EFRF上,以XML的形式输出软件功能需求结构化表达并存储在系统中。 [0006] The present invention addresses the above shortcomings of the prior art, to provide an automatic extraction system based on the functional requirements described frame expanded functional requirements, the functional requirements document for the functional requirement document based analysis as an input the parser, through a series of predefined conversion rules for parsing the results of the conversion, mapped onto EFRF, output software functional requirements structured as XML expressed and stored in the system.

[0007] 本发明是通过以下技术方案实现的,本发明包括:语法分析模块和功能需求转换模块,其中:语法分析模块基于语法分析器,功能需求转换模块根据预定义的转换规则将语法分析器的分析结果映射到EFRF上,通过两个模块的分析和转换。 [0007] The present invention is achieved by the following technical solutions, the present invention includes: parsing module function requirements and conversion module, wherein: the parser module is based on the syntax analysis, functional requirements parser conversion module according to predefined conversion rule the results mapped onto EFRF, by two modules analyzed and transformed.

[0008] 所述的EFRF包括10个功能需求的描述维度,每个功能需求都可以用一个EFRF中的这10个维度来描述。 [0008] 10 comprises the EFRF dimensions described functional requirements, the functional requirements of each can be used in which a EFRF 10 dimensions is described.

[0009] 本发明的原理是:针对软件功能需求,其上下文可以用所定义的EFRF框架来描述,该框架包括10个描述维度,即=Agentive,功能内容的发起者;Action,功能描述的行为;Objective,功能行为的作用对象;Agentmod,功能发起者的约束;Objmod,功能作用对象的约束;Locational,功能相关的地点;Temporal,功能相关的时间,包括发生频率,持续时间等;Manner,功能实现的方式,包括工具,条件等;Goal,功能的目标;Constraint,功能实现的其他限制条件。 [0009] The principle of the invention is: the demand for software function, which can be described by the context defined EFRF frame, the frame 10 includes a description of the dimensions, i.e. = Agentive, function initiator content; the Action behavior, the functions described ; role of the object in Objective, functional behavior; constraint Agentmod, function initiator; constraint Objmod, functional role object; Locational, functions related to location; Temporal, functions related to time, including frequency, duration and the like; by Manner, function manner, including the tools and conditions; target goal, function; constraint, other restrictions function implemented. 进一步地,经观察和实现发现,上述软件功能需求的描述维度都与功能需求的语法分析结果具有关联性,可以依据一定的转换规则实现自动抽取。 Further, it was observed and found to achieve, the above described dimensions are software functional requirements and functional requirements of the results of parsing relevant, it can be automatically extracted according to a certain conversion rule. 因此,本发明借助于语法分析器和一组定义的转换规则实现了软件功能需求的自动抽取。 Accordingly, the present invention is by means of a parser and a set of transformation rules defined by the software to achieve the functional requirements of the automatic extraction.

[0010] 本发明有益的效果是:提出了基于EFRF的软件功能需求的自动抽取方法并实现了相应系统;借助于所构建的语法分析器,提高了对软件功能需求文档进行自动化的分析和处理的能力;基于所构建的EFRF描述框架和转换规则,语法分析器的分析结果被映射到一个多维度描述的软件功能需求的结构化模型,从而提高了软件功能需求描述的全面性和准确性。 [0010] Advantageous effects of the present invention is: based on the proposed method of automatic extraction software function EFRF demand and to achieve a corresponding system; constructed by means of a parser, to improve the analysis and processing software functional requirements document automation capacity; structural model EFRF described conversion rule framework and constructed, the parser analysis result is mapped to a multi-dimensional software described functional requirements, thereby improving the accuracy and comprehensiveness of the software functions described requirements. 此外,本发明所构建的系统,可以分析所有符合IEEE-STD-830标准的软件需求文档,能大大降低在软件功能需求分析中所涉及的手工劳动,从而为企业节省了人力物力,为企业软件的大规模定制提供了帮助,在企业软件开发上具有很高的应用和商业价值。 In addition, the construction of the present invention, the system can analyze all the software requirements document compliance with IEEE-STD-830 standards, can greatly reduce manual labor in the software functional requirements analysis involved, so as to enterprises save manpower and resources for enterprise software mass customization has helped, with a high commercial value and applications in the enterprise software development.

附图说明 BRIEF DESCRIPTION

[0011] 图1是本发明的系统框架图。 [0011] FIG. 1 is a system frame according to the present invention.

[0012] 图2是语法结构分析结果示例。 [0012] FIG. 2 is an example of a syntax structure analysis results.

[0013] 图3是语法依存分析结果示例。 [0013] FIG. 3 is an example of the analysis result of the syntax dependent.

[0014] 图4是系统最终生成的功能需求模型示例。 [0014] FIG. 4 is an example of the functional needs of the system model finally generated.

具体实施方式 Detailed ways

[0015] 下面结合附图对本发明的实施例作详细说明,本实施例在以本发明技术方案为前提下进行实施,给出了详细的实施方式和具体的操作过程,但本发明的保护范围不限于下述的实施例。 [0015] The following embodiments in conjunction with the accompanying drawings of embodiments of the present invention will be described in detail, in the present embodiment to embodiments of the present invention, the technical solution under the premise that, given the specific operation and detailed embodiments, but the scope of the present invention It is not limited to the following examples.

[0016] 如图1所示,本实施例包括:语法分析模块和功能需求转换模块,其中:语法分析模块基于语法分析器,功能需求转换模块根据预定义的转换规则将语法分析器的分析结果映射到EFRF上,通过两个模块的分析和转换。 [0016] 1, the present embodiment includes: parsing module function requirements and conversion module, wherein: based parser module will analyze the functional requirements into a parser module parsing according to predefined conversion rule results mapped to EFRF, by two modules analyzed and transformed.

[0017] 所述的EFRF包括10个功能需求的描述维度,每个功能需求都可以用一个EFRF中的这10个维度来描述。 EFRF according to [0017] 10 comprising a functional dimension requirements described, each functional requirements can be used in which a EFRF 10 dimensions is described.

[0018] 该实施例的输入是一组现有软件系统的功能需求文档,所有文档为符合IEEE-STD-830标准的需求文档。 [0018] Example embodiment of the input is a set of functional requirements document existing software systems, all documents to comply with IEEE-STD-830 standard requirements document. 所采用的实施例是由249个句子组成的功能需求描述,包含5,669个词,不含标点的单词为5,189个。 Example embodiments are employed by the functional requirements 249 sentences description contains 5,669 words, without punctuation one word is 5,189.

[0019] 系统第一个主模块是语法分析器,基于斯坦福语法分析器Manford Parser和文本工程框架平台GATE开发实现语法分析器,并将上述功能需求文档输入到语法分析器中进行处理。 [0019] System main module is a parser, based on the Stanford and text parser Manford Parser GATE internet engineering development frame parser implemented, and inputs to the functional requirements document parser for processing. 语法分析器的实现部分包括:1)分词与词性标注模块,对需求文本进行分词、词性标注;2)语法结构树分析模块,对经过分词和词性标注的需求文本进行语法结构分析, 分析出句子中的所有结构信息;3)依存关系分析模块,通过这个模块,得到的是句子内各成分之间的依存关系,这里的依存关系指的是二元关系,比如直接宾语,补语等关系。 Achieve some parsers comprises: 1) word and POS tagging module, the need for text segmentation, POS tagging; 2) syntax tree analysis module, the elapsed word and POS tagging needs text syntax structure analysis, the sentence All configuration information; 3) dependency analysis module, through this module, to obtain the interdependence between the various components within a sentence, where dependencies refers binary relation, such as direct object relationship, complement and the like.

[0020] 本实施例通过将Manford Parser整合到GATE框架下,采用了两个步骤来实现语法分析器。 [0020] In this embodiment, the integrated GATE Manford Parser to the lower framework, the two steps to achieve the parser. 第一步,调用Manford Parser,将解析结果的依存关系加入到GATE下的文档中作为iToken的属性。 The first step, calling Manford Parser, the analytical results of the dependencies added to the document in GATE as iToken property. 从Manford Parser处理之后的文档中除了标注词性信息之外还标注了句子内部成分的依存关系属性。 From the document after addition Manford Parser process further information other than speech marked marked dependence property of internal components of the sentence. 对依存关系解析的最小单元是一个句子,生成的是一个句子内部各个成分之间的依存关系。 The minimum unit dependency analysis of a sentence is generated dependencies between the various internal components of a sentence. 第二步,扩展GATE中的NE模块,利用分析好的语义角色,将其与功能需求本体中的实体概念相结合,利用之前定义好的一系列转换规则,将这套规则用GATE可识别的符合JAPE语法规则表达,从而实现EFRF实体的识别。 The second step, the extended GATE NE module using the semantic analysis of good character, which is combined with the concept of functional entities requirements body, a series of transformation rules defined before use, this set of rules can be used to identify the GATE JAPE expression in line with the rules of grammar, in order to achieve recognition EFRF entity. 因为在命名实体识别的时候的依据是依存关系,所以,在JAPE规则中很重要的一个输入就是D印endency。 Because according to the time named entity recognition that dependency, so it is important in JAPE rule in a D input is printed endency. 这里Lookup,Token, Dependency都是解析之后的词的属性,将这些属性的值作为规则的判别标准。 Here Lookup, Token, after the Dependency attribute words are parsed, the values ​​of these properties as a criterion rules.

[0021] 系统第二个主模块是功能需求转换器,基于所提出的EFRF框架和预定义的转换规则,对语法分析结果进行再处理,从而将其映射到软件功能需求的EFRF框架上。 [0021] The second system is a functional master module needs converters, and based on EFRF frame of predefined conversion rules proposed syntax analysis result reprocessing to be mapped to the functional requirements of software EFRF frame. 功能需求转换器的实现部分是一个EFRF命名实体识别模块,通过事先定义的转换规则最终实现自动分析和映射。 Achieve some functional requirements EFRF converter is a named entity recognition module, and ultimately through the automatic analysis and mapping conversion rule defined in advance.

[0022] 本实施例通过定义有效的转换规则,将语法分析结果映射到功能需求描述框架EFRF下,实现了功能需求转换器。 [0022] In this embodiment, define a valid conversion rule mapping syntax analysis result to the functional requirements described frame EFRF, the converter of the functional requirements. 表1是其中的部分转换规则,本实施例对功能需求中的主动和被动语态分别进行分析,针对不同的语态建立不同的规则。 Table 1 is a part of the conversion rule, according to the present embodiment, the functional requirements of the active and passive analysis, respectively, to establish different rules for different voice. 表格第一列是要匹配的实体概念,与EFRF中的维度一一对应,表示当规则匹配了之后,该文本片段将要加注的标签名称,也就是概念名称。 Row of the table is a conceptual entities to match with the dimensions EFRF correspond, when expressed tag name matches the rule, the text segment is to be raised, that is, the name of the concept. 第二列和第三列分别是主动语态和被动语态下的转换规则。 The second and third columns are the active voice conversion rules and passive. 规则中出现的符号的含义如下。 The meaning of the rules that appear symbols are as follows. 规则中的概念名称,如“Action” :指的是已经被其他规则识别出来并加了标注的“Action”实体。 The concept of the rule name, such as "Action": refers to the other rules have been identified and added it marked "Action" entity. 规则左括号前面的名称,如“Subj”:指的是通过自然语言处理得到的依存关系。 Left parenthesis in front of the name of the rule, such as "Subj": refers to the dependency obtained by natural language processing. 如果两个文本片段之间的关系是Subj,则利用这条规则进行匹配,看是否满足规则的其他条件。 If the relationship between the two is the Subj text segments, using the matching rule, to see if the other conditions of the rule is satisfied. 规则“obj (Action, Χ) ”表示的是,当识别出了Action之后,而且在文本中有一个短语与Action之间的依存关系是obj (包含obj的各个子关系), 那么这个短语就会被标记为Objective。 Rules "obj (Action, Χ)" represent that, when identified the Action, but also in the text dependency between a phrase and Action is a obj (each containing child relationship of obj), then this phrase will It is marked as Objective. 规则“conj (objective, X) ”表示的是,当识别出了Objective,而且在文本中有一个短语与Objective之间的依存关系是con_j (包括con_j的各个子关系),那么这个短语也同样被标记为Objective。 Rule "conj (objective, X)" indicates that, when the Objective identified, but also in the text of a dependency relationship between the phrases are con_j Objective (including the respective sub con_j relationship), then the phrase is similarly marked Objective. 最后根据识别出来的所有Objective的语序进行排序,得到一个连续的文本片段或者几个独立的短语。 Finally, sorted according to the word order all Objective identified to obtain a continuous or several separate text segment phrases.

[0023] 实施例的工作过程: [0023] The operation of this embodiment:

[0024] 第一步,用户向该抽取系统发出如下请求: [0024] The first step, the user issues a request to the extraction system as follows:

[0025] Teacher assistant mark students' homework in the lab at 7' clock. [0025] Teacher assistant mark students 'homework in the lab at 7' clock.

[0026] 该请求是目标应用需要满足的一个功能需求的文本描述。 [0026] The request is a text description of the function of the target application requirements need to be met.

[0027] 第二步,上述文本描述通过语法分析器的词性标注模块进行分析。 [0027] The second step, the above described analysis by the text parser speech tagging module. 对上述文本需求进行分词与词性标注,输出结果如下所示:[0028] Teacher/NN assistant/NN mark/VBP students/NNS' /POS homework/NN in/IN the/DTlab/NN at/IN T clock/NN. /. Of said text needs word and POS tagging, the output results are as follows: [0028] Teacher / NN assistant / NN mark / VBP students / NNS '/ POS homework / NN in / IN the / DTlab / NN at / IN T clock / NN. /.

[0029] 在上面结果中,每个单词紧跟的“/”后面的符号表示分析结果中该单词的词性,例如assistant/NN表示assistant的词性是名词。 [0029] In the above results, each of the word immediately following "/" symbol indicates the rear part of speech of the word analysis results, e.g. assistant / NN represents the part of speech is a noun assistant.

[0030] 第三步,使用上述分词与词性分析结果,通过语法分析器对其进行语法结构树解析,分析出句子中的所有结构信息。 [0030] The third step, by using the word and part of speech analysis, syntax parser tree by its deconvolution analysis, all the configuration information in the sentence. 例如“(NP(NN Teacher) (NN assistant)) ”表示Treacher assistant是由两个名词组成的一个名词短语(Noun Phrase)。 For example "(NP (NN Teacher) (NN assistant))" represents Treacher assistant is a noun phrase consisting of two nouns (Noun Phrase). 对此示例的语法结构树展示请见图2。 For this example the syntax tree display See Figure 2.

[0031] 第四步,基于上述语法结构树信息,使用语法分析器对实施例进行语法依存关系解析,分析出句子中的直接/间接主语、直接/间接宾语、补语、修饰语等等语法上的依存关系对。 [0031] The fourth step, based on the syntax tree information, using parser syntax Example dependency analysis, the analysis of the sentence direct / indirect subject, direct / indirect object, complement, and the like syntax modifier dependencies right. 例如:dobj(mark-3,homework-6)表示mark 的直接宾语(direct object)是homework。 For example: dobj (mark-3, homework-6) represents a direct object of the mark (direct object) is homework. 上述需求文本的语法依存对展示请见图3。 Syntax dependency of these requirements for text display See Figure 3.

[0032] 第五步,将上述获得的语法依存对输入到功能需求转换器中,利用表1定义的转换规则将语法依存对中识别的语法成分对应到相应地EFRF描述维度上。 [0032] The fifth step, the dependency grammar obtained above is input to the functional requirements of the converter, using the conversion rules defined in Table 1 will be dependent on the identified grammar syntax corresponds to the component described respectively EFRF dimension. 例如, ηsubj (mark, assistant) ψ "mark" I^iKSiJ^ Action,胃“assistant” I^iR另0¾ Agentive。 For example, ηsubj (mark, assistant) ψ "mark" I ^ iKSiJ ^ Action, stomach "assistant" I ^ iR another 0¾ Agentive.

[0033] 表1( ?-指的是任意一个依存关系;* -指的是任意长度的字符;X-指的是当规则匹配之后需要加注标签的文本片段;&_指的是前后两条规则应该同时满足) [0033] Table 1 (- refers to any one dependencies;? * - refers to the character of any length; X-refers to the need to labeling after when the rule matches the text segments; & _ refers to the two front rule should meet)

Figure CN102147731AD00071

[0035] 然而,某些语法成分可能会被识别到多个描述维度上。 [0035] However, some components may be identified grammar to multiple dimensions is described. 例如,根据表1定义的规贝prep_in (homework, lab)中“lab,,可能会被识另Ij为Locational, Temporal 或Manner 等。为了避免冲突,进一步地判定这三个由介词短语构成的语法依存对中相应成分的语义类型。 For example, according to the rules defined in Table 1 prep_in shellfish (homework, lab) in "lab ,, Ij may be recognized as another like Locational, Temporal or Manner. To avoid conflicts, these three grammar is further determined by the configuration of prepositional phrases dependent on the respective semantic type components.

[0036] 1)针对Locational维度,定义了如下的词表(可根据需要进一步扩充)表示某个词的语义类型: [0036] 1) For Locational dimension, defines the following vocabulary (may be further expanded as needed) represent a word semantic type:

[0037] PLACE :home, lab, office, any place, bus stop, station, train station, crossroad,... [0037] PLACE: home, lab, office, any place, bus stop, station, train station, crossroad, ...

[0038] PLACENAME:Asia, Europe, Africa, Shanghai, Beijing,... [0038] PLACENAME: Asia, Europe, Africa, Shanghai, Beijing, ...

[0039] 2)针对"Temporal维度,定义了如下的词表: [0039] 2) for "Temporal dimension, we define the following vocabulary:

[0040] TIME :morning,noon,afternoon, evening, night, midnight, Number,clock,... [0040] TIME: morning, noon, afternoon, evening, night, midnight, Number, clock, ...

[0041] DATE :Monday,Tuesday, Wednesday, January, February,... [0041] DATE: Monday, Tuesday, Wednesday, January, February, ...

[0042] DURATION :Number Days/Months/Years... [0042] DURATION: Number Days / Months / Years ...

[0043] FREQUENCY :once,Number Times,...[0044] 3)针对Manner维度,采用“排除法”,即相应的介词短语所引导的语法成分不属于Locational和^Temporal定义的词表中,即可判定为Manner。 [0043] FREQUENCY: once, Number Times, ... [0044] 3) Manner for dimensions, the use of "exclusion", i.e. corresponding prepositional guided grammatical elements and not Locational ^ Temporal defined vocabulary, It can be judged Manner.

[0045] 根据上述定义的词表,语法依存对中的相应成分被划分到某个语义类型上,从而进一步判定其所属的维度。 [0045] According to the above-defined vocabulary, syntax dependent on the respective components are divided into a semantic types, thus further dimension to which it belongs is determined. 例如,在上述实施例中,PRPjn(homework,lab),由于lab属于PLACE语义类型,因而“in the lab”被判定为Locational维度。 For example, in the above embodiment, PRPjn (homework, lab), since the lab belonging PLACE semantic types, so "in the lab" is determined Locational dimension. 再如,prep_at (lab, 7,clock),由于7,clock属于TIME语义类型,因而“at 7,clock”被判定为Temporal维度。 Again, prep_at (lab, 7, clock), due 7, clock belonging TIME semantic types, so "at 7, clock" Temporal dimension is determined.

[0046] 图4展示了上述请求经系统处理后的最终结果。 [0046] FIG. 4 shows the final result of the request processed by the system. 可以看到,在整个界面的右边的地方是扩展的要抽取的十种EFRF语义命名实体(Action,Agentive, Agentmod, Constraint, Goal, Locational, Manner, Objective, Objmod, Temporal)。 It can be seen in the right places throughout the interface is extended to ten EFRF semantic extraction of named entities (Action, Agentive, Agentmod, Constraint, Goal, Locational, Manner, Objective, Objmod, Temporal). 抽取出来的功能性需求由10个维度描述,这个EFRF框架反映了功能需求描述中的语义特征,因此,它能作为一个结构化框架,来帮助进一步全面地分析功能需求。 Extracted functional requirements described by the 10 dimensions, this reflects a semantic feature EFRF framework described in functional requirements, therefore, it can as a structural framework to facilitate further comprehensive analysis of functional requirements.

[0047] 表2是通过运行系统对实验数据进行分析和评价后得到的结果。 [0047] Table 2 is the analysis and evaluation of experimental results obtained by running the system.

[0048] 表2系统抽取效果的评测结果 [0048] The evaluation results in Table 2, the effect of extraction system

[0049] [0049]

Figure CN102147731AD00081

[0050] 实验结果显示了准确率普遍比召回率高,这是由基于规则的信息抽取的特点所决定的。 [0050] The experimental results show generally higher than the accuracy of the recall rate, which is determined by the information extraction rule-based features. 其中Agentive,Action, Objective, Agentmod和Objmod的抽取效果在准确率和召回率上都较高。 Wherein Agentive, extraction effect Action, Objective, Agentmod Objmod and on precision and recall rates are high. 但是对于其他的语义成分,基于规则的方法得到的召回率相对较低。 But for other semantic components, rule-based method to obtain recall rate is relatively low. 但总体上,平均F值能达到72.6%,说明该方法以及形成的系统是比较有效的。 But in general, the average value of F can reach 72.6%, indicating that the method and system is more effective formed. 本实例结合Stanford Parser,通过引入EFRF模型和定制好的转换规则,实现了一个可以实际应用的系统。 Examples of this binding Stanford Parser, and by introducing EFRF model-customized conversion rule, a system can achieve practical application. 对实验数据的测试结果表明了该系统的有效性。 The test results of the experimental data show the effectiveness of the system.

Claims (2)

1.一种基于扩展功能需求描述框架的功能需求自动抽取系统,其特征在于,包括:语法分析模块和功能需求转换模块,其中:语法分析模块基于语法分析器,功能需求转换模块根据预定义的转换规则将语法分析器的分析结果映射到EFRF上,通过两个模块的分析和转换。 An automatic extraction system based on the functional requirements described demand extension frame, characterized in that, comprising: a parsing module function requirements and module conversion, wherein: the parser module is based on parsing, conversion module according to the functional requirements of a predefined the analysis result of the conversion rule parser mapped onto EFRF, by two modules analyzed and transformed.
2.根据权利要求1所述的基于扩展功能需求描述框架的功能需求自动抽取系统,其特征是,所述的EFRF包括10个功能需求的描述维度,每个功能需求都可以用一个EFRF中的这10个维度来描述。 2. The automatic extraction functional requirements based on the extended functional requirements described in the frame system of claim 1, wherein said EFRF comprises 10 functional requirements described dimensions, each functional requirements can be used in a EFRF these 10 dimensions to describe.
CN 201110099316 2011-04-20 2011-04-20 Automatic functional requirement extraction system based on extended functional requirement description framework CN102147731A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 201110099316 CN102147731A (en) 2011-04-20 2011-04-20 Automatic functional requirement extraction system based on extended functional requirement description framework

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 201110099316 CN102147731A (en) 2011-04-20 2011-04-20 Automatic functional requirement extraction system based on extended functional requirement description framework

Publications (1)

Publication Number Publication Date
CN102147731A true true CN102147731A (en) 2011-08-10

Family

ID=44422008

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 201110099316 CN102147731A (en) 2011-04-20 2011-04-20 Automatic functional requirement extraction system based on extended functional requirement description framework

Country Status (1)

Country Link
CN (1) CN102147731A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103294653A (en) * 2012-02-22 2013-09-11 通用汽车环球科技运作有限责任公司 System and method for generating high-quality formal executable software feature requirements

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1542736A (en) * 2003-05-01 2004-11-03 微软公司 Rules-based grammar for slots and statistical model for preterminals in natural language understanding system
US20100275179A1 (en) * 2009-04-23 2010-10-28 International Business Machines Corporation Extracting a system modelling meta-model language model for a system from a natural language specification of the system

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1542736A (en) * 2003-05-01 2004-11-03 微软公司 Rules-based grammar for slots and statistical model for preterminals in natural language understanding system
US20100275179A1 (en) * 2009-04-23 2010-10-28 International Business Machines Corporation Extracting a system modelling meta-model language model for a system from a natural language specification of the system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
木云鹤: "《简历及软件需求信息抽取方法研究》", 《中国优秀硕士学位论文全文数据库》, 26 July 2010 (2010-07-26) *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103294653A (en) * 2012-02-22 2013-09-11 通用汽车环球科技运作有限责任公司 System and method for generating high-quality formal executable software feature requirements
CN103294653B (en) * 2012-02-22 2016-12-28 通用汽车环球科技运作有限责任公司 System and method for official executable software features for producing high quality requirements

Similar Documents

Publication Publication Date Title
Diab Second generation AMIRA tools for Arabic processing: Fast and robust tokenization, POS tagging, and base phrase chunking
US20070136698A1 (en) Method, system and apparatus for a parser for use in the processing of structured documents
Corley et al. Finding Syntactic Structure in Unparsed Corpora The Gsearch Corpus Query System
Ide et al. The manually annotated sub-corpus: A community resource for and by the people
Watt The parsing problem for affix grammars
Padó et al. Optimal constituent alignment with edge covers for semantic projection
US20130151238A1 (en) Generation of Natural Language Processing Model for an Information Domain
US7512633B2 (en) Conversion of hierarchically-structured HL7 specifications to relational databases
CN103049435A (en) Text fine granularity sentiment analysis method and text fine granularity sentiment analysis device
CN101079031A (en) Web page subject extraction system and method
CN102279894A (en) Find a semantic-based, integrated approach and provide comments and information search systems
US8561014B2 (en) Extracting a system modelling meta-model language model for a system from a natural language specification of the system
CN101976188A (en) OpenApi data automatic loading system oriented to AJAX protocol
CN102890702A (en) Internet forum-oriented opinion leader mining method
CN104408078A (en) Construction method for key word-based Chinese-English bilingual parallel corpora
US20110040553A1 (en) Natural language processing
CN101446942A (en) Semantic character labeling method of natural language sentence
US20120078950A1 (en) Techniques for Extracting Unstructured Data
Sarkar et al. Some experiments on indicators of parsing complexity for lexicalized grammars
JP2003167898A (en) Information retrieving system
CN102866989A (en) Viewpoint extracting method based on word dependence relationship
Furlan et al. Semantic similarity of short texts in languages with a deficient natural language processing support
CN102937949A (en) Method and system for checking English spelling in rich text editor
CN101488126A (en) Double-language sentence alignment method and device
CN104268200A (en) Unsupervised named entity semantic disambiguation method based on deep learning

Legal Events

Date Code Title Description
C06 Publication
C10 Entry into substantive examination
C12 Rejection of a patent application after its publication