1,592
个编辑
更改
跳到导航
跳到搜索
Moved page from wikipedia:en:Information extraction (history)
此词条暂由彩云小译翻译，翻译字数共1119，未经人工整理和审校，带来阅读不便，请见谅。
{{short description|Automatically extracting structured information from un- or semi-structured machine-readable documents, such as human language texts}}
'''Information extraction''' ('''IE''') is the task of automatically extracting structured information from [[unstructured data|unstructured]] and/or semi-structured [[machine-readable data|machine-readable]] documents and other electronically represented sources. In most of the cases this activity concerns processing human language texts by means of [[natural language processing]] (NLP). Recent activities in [[multimedia]] document processing like automatic annotation and content extraction out of images/audio/video/documents could be seen as information extraction
Information extraction (IE) is the task of automatically extracting structured information from unstructured and/or semi-structured machine-readable documents and other electronically represented sources. In most of the cases this activity concerns processing human language texts by means of natural language processing (NLP). Recent activities in multimedia document processing like automatic annotation and content extraction out of images/audio/video/documents could be seen as information extraction
信息抽取是从非结构化和/或半结构化的机器可读文档和其他电子表示的源中自动提取结构化信息的任务。在大多数情况下，这种活动涉及到通过自然语言处理(NLP)来处理人类语言文本。最近在多媒体文档处理方面的活动，如图像/音频/视频/文档的自动注释和内容提取，可以被视为信息抽取
Due to the difficulty of the problem, current approaches to IE focus on narrowly restricted domains. An example is the extraction from newswire reports of corporate mergers, such as denoted by the formal relation:
Due to the difficulty of the problem, current approaches to IE focus on narrowly restricted domains. An example is the extraction from newswire reports of corporate mergers, such as denoted by the formal relation:
由于这个问题的难度，目前的 IE 方法集中在狭窄的限制领域。一个例子是从新闻通讯社关于公司合并的报告中摘录，例如用正式关系表示:
:<math>\mathrm{MergerBetween}(company_1, company_2, date)</math>,
<math>\mathrm{MergerBetween}(company_1, company_2, date)</math>,
公司1，公司2，日期,
from an online news sentence such as:
from an online news sentence such as:
从一个在线新闻句，如:
:''"Yesterday, New York based Foo Inc. announced their acquisition of Bar Corp."''
"Yesterday, New York based Foo Inc. announced their acquisition of Bar Corp."
昨天，总部位于纽约的 Foo 公司宣布收购 Bar corp.
A broad goal of IE is to allow computation to be done on the previously unstructured data. A more specific goal is to allow [[logical reasoning]] to draw inferences based on the logical content of the input data. Structured data is semantically well-defined data from a chosen target domain, interpreted with respect to category and [[context (language use)|context]].
A broad goal of IE is to allow computation to be done on the previously unstructured data. A more specific goal is to allow logical reasoning to draw inferences based on the logical content of the input data. Structured data is semantically well-defined data from a chosen target domain, interpreted with respect to category and context.
IE 的一个广泛的目标是允许在之前的非结构化数据上进行计算。一个更具体的目标是允许逻辑推理基于输入数据的逻辑内容做出推论。结构化数据是来自选定目标域的语义上定义良好的数据，根据类别和上下文进行解释。
Information Extraction is the part of a greater puzzle which deals with the problem of devising automatic methods for text management, beyond its transmission, storage and display. The discipline of [[information retrieval]] (IR)<ref>{{Cite journal|url = http://www.cs.bilkent.edu.tr/~guvenir/courses/CS550/Seminar/freitag2000-ml.pdf|title = Machine Learning for Information Extraction in Informal Domains|last = FREITAG|first = DAYNE|journal = 2000 Kluwer Academic Publishers. Printed in the Netherlands}}</ref> has developed automatic methods, typically of a statistical flavor, for indexing large document collections and classifying documents. Another complementary approach is that of [[natural language processing]] (NLP) which has solved the problem of modelling human language processing with considerable success when taking into account the magnitude of the task. In terms of both difficulty and emphasis, IE deals with tasks in between both IR and NLP. In terms of input, IE assumes the existence of a set of documents in which each document follows a template, i.e. describes one or more entities or events in a manner that is similar to those in other documents but differing in the details. An example, consider a group of newswire articles on Latin American terrorism with each article presumed to be based upon one or more terroristic acts. We also define for any given IE task a template, which is a(or a set of) case frame(s) to hold the information contained in a single document. For the terrorism example, a template would have slots corresponding to the perpetrator, victim, and weapon of the terroristic act, and the date on which the event happened. An IE system for this problem is required to “understand” an attack article only enough to find data corresponding to the slots in this template.
Information Extraction is the part of a greater puzzle which deals with the problem of devising automatic methods for text management, beyond its transmission, storage and display. The discipline of information retrieval (IR) has developed automatic methods, typically of a statistical flavor, for indexing large document collections and classifying documents. Another complementary approach is that of natural language processing (NLP) which has solved the problem of modelling human language processing with considerable success when taking into account the magnitude of the task. In terms of both difficulty and emphasis, IE deals with tasks in between both IR and NLP. In terms of input, IE assumes the existence of a set of documents in which each document follows a template, i.e. describes one or more entities or events in a manner that is similar to those in other documents but differing in the details. An example, consider a group of newswire articles on Latin American terrorism with each article presumed to be based upon one or more terroristic acts. We also define for any given IE task a template, which is a(or a set of) case frame(s) to hold the information contained in a single document. For the terrorism example, a template would have slots corresponding to the perpetrator, victim, and weapon of the terroristic act, and the date on which the event happened. An IE system for this problem is required to “understand” an attack article only enough to find data corresponding to the slots in this template.
信息抽取是一个更大的难题的一部分，它涉及的问题是设计文本管理的自动方法，超越了它的传输，存储和显示。信息检索学科已经开发出了自动化的方法，典型的统计方法，用于为大型文档集合建立索引和对文档进行分类。另一个互补的方法是自然语言处理(NLP) ，它解决了人类语言处理建模的问题，在考虑到任务的规模时取得了相当大的成功。就难度和重点而言，IE 处理介于 IR 和 NLP 之间的任务。在输入方面，IE 假定存在一组文档，其中每个文档都遵循一个模板，即。以类似于其他文档中的方式描述一个或多个实体或事件，但在细节上有所不同。例如，考虑一组关于拉丁美洲恐怖主义的新闻专线文章，每一条都被认为是基于一种或多种恐怖主义行为。我们还为任何给定的 IE 任务定义了一个模板，它是一个(或一组)案例框架，用于保存单个文档中包含的信息。对于恐怖主义的例子，一个模板应该有与恐怖主义行为的肇事者、受害者和武器相对应的位置，以及事件发生的日期。针对这个问题的 IE 系统需要“理解”一篇攻击文章，只要找到与此模板中插槽相对应的数据即可。
==History==
Information extraction dates back to the late 1970s in the early days of NLP.<ref>{{cite book|chapter-url=https://www.aclweb.org/anthology/A92-1024|chapter=Automatic Extraction of Facts from Press Releases to Generate News Stories|last1=Andersen|first1=Peggy M.|last2=Hayes|first2=Philip J.|citeseerx=10.1.1.14.7943|last3=Huettner|first3=Alison K.|last4=Schmandt|first4=Linda M.|last5=Nirenburg|first5=Irene B.|last6=Weinstein|first6=Steven P.|title=Proceedings of the third conference on Applied natural language processing -|year=1992|pages=170–177|doi=10.3115/974499.974531|s2cid=14746386}}</ref> An early commercial system from the mid-1980s was JASPER built for [[Reuters]] by the Carnegie Group Inc with the aim of providing [[Electronic communication network|real-time financial news]] to financial traders.<ref>{{cite book|url=http://pdfs.semanticscholar.org/2c90/fa59c6d9beed8dcb0e844725b872d3f33a35.pdf|archive-url=https://web.archive.org/web/20190220184608/http://pdfs.semanticscholar.org/2c90/fa59c6d9beed8dcb0e844725b872d3f33a35.pdf|url-status=dead|archive-date=2019-02-20|title=Information Extraction|last1=Cowie|first1=Jim|last2=Wilks|first2=Yorick|page=3|year=1996|citeseerx=10.1.1.61.6480|s2cid=10237124}}</ref>
Information extraction dates back to the late 1970s in the early days of NLP. An early commercial system from the mid-1980s was JASPER built for Reuters by the Carnegie Group Inc with the aim of providing real-time financial news to financial traders.
信息抽取可以追溯到20世纪70年代末 NLP 的早期。早期的商业系统是20世纪80年代中期由卡内基集团公司为路透社建立的 JASPER 系统，其目的是为金融交易员提供实时的财经新闻。
Beginning in 1987, IE was spurred by a series of [[Message Understanding Conference]]s. MUC is a competition-based conference<ref>Marco Costantino, Paolo Coletti, Information Extraction in Finance, Wit Press, 2008. {{ISBN|978-1-84564-146-7}}</ref> that focused on the following domains:
Beginning in 1987, IE was spurred by a series of Message Understanding Conferences. MUC is a competition-based conference that focused on the following domains:
从1987年开始，IE 受到了一系列信息理解会议的激励。MUC 是一个以竞争为基础的会议，重点关注以下领域:
*MUC-1 (1987), MUC-2 (1989): Naval operations messages.
*MUC-3 (1991), MUC-4 (1992): Terrorism in Latin American countries.
*MUC-5 (1993): [[Joint venture]]s and microelectronics domain.
*MUC-6 (1995): News articles on management changes.
*MUC-7 (1998): Satellite launch reports.
Considerable support came from the U.S. Defense Advanced Research Projects Agency ([[DARPA]]), who wished to automate mundane tasks performed by government analysts, such as scanning newspapers for possible links to terrorism.{{citation needed|date=March 2017}}
Considerable support came from the U.S. Defense Advanced Research Projects Agency (DARPA), who wished to automate mundane tasks performed by government analysts, such as scanning newspapers for possible links to terrorism.
美国国防部高级研究计划局(DARPA)提供了大量的支持，他们希望将政府分析人员执行的日常任务自动化，比如扫描报纸以寻找与恐怖主义的可能联系。
==Present significance==
The present significance of IE pertains to the growing amount of information available in unstructured form. [[Tim Berners-Lee]], inventor of the [[world wide web]], refers to the existing [[Internet]] as the web of ''documents'' <ref>{{cite web|url=http://tomheath.com/papers/bizer-heath-berners-lee-ijswis-linked-data.pdf|title=Linked Data - The Story So Far}}</ref> and advocates that more of the content be made available as a [[semantic web|web of ''data'']].<ref>{{cite web|url=http://www.ted.com/talks/tim_berners_lee_on_the_next_web.html|title=Tim Berners-Lee on the next Web}}</ref> Until this transpires, the web largely consists of unstructured documents lacking semantic [[metadata]]. Knowledge contained within these documents can be made more accessible for machine processing by means of transformation into [[relational database|relational form]], or by marking-up with [[XML]] tags. An intelligent agent monitoring a news data feed requires IE to transform unstructured data into something that can be reasoned with. A typical application of IE is to scan a set of documents written in a [[natural language]] and populate a database with the information extracted.<ref>[[Rohini Kesavan Srihari|R. K. Srihari]], W. Li, C. Niu and T. Cornell,"InfoXtract: A Customizable Intermediate Level Information Extraction Engine",[https://web.archive.org/web/20080507153920/http://journals.cambridge.org/action/displayIssue?iid=359643 Journal of Natural Language Engineering],{{dead link|date=September 2020}} Cambridge U. Press, 14(1), 2008, pp.33-69.</ref>
The present significance of IE pertains to the growing amount of information available in unstructured form. Tim Berners-Lee, inventor of the world wide web, refers to the existing Internet as the web of documents and advocates that more of the content be made available as a web of data. Until this transpires, the web largely consists of unstructured documents lacking semantic metadata. Knowledge contained within these documents can be made more accessible for machine processing by means of transformation into relational form, or by marking-up with XML tags. An intelligent agent monitoring a news data feed requires IE to transform unstructured data into something that can be reasoned with. A typical application of IE is to scan a set of documents written in a natural language and populate a database with the information extracted.
IE 目前的重要意义在于以非结构化的形式获得越来越多的信息。万维网的发明者 Tim Berners-Lee 将现有的互联网称为文档网络，并主张更多的内容以数据网络的形式提供。在此之前，网络大部分是由缺乏语义元数据的非结构化文档组成的。这些文档中包含的知识可以通过转换为关系形式或使用 XML 标记使机器处理更容易访问。一个监控新闻数据源的智能代理需要 IE 将非结构化数据变成可以理解的东西。IE 的典型应用程序是扫描一组用自然语言编写的文档，并用提取的信息填充数据库。
==Tasks and subtasks==
Applying information extraction to text is linked to the problem of [[text simplification]] in order to create a structured view of the information present in free text. The overall goal being to create a more easily machine-readable text to process the sentences. Typical IE tasks and subtasks include:
Applying information extraction to text is linked to the problem of text simplification in order to create a structured view of the information present in free text. The overall goal being to create a more easily machine-readable text to process the sentences. Typical IE tasks and subtasks include:
将信息抽取应用于文本是与文本简化问题联系在一起的，以便创建一个自由文本信息的结构化视图。总体目标是创建一个更容易机器阅读的文本来处理句子。典型的 IE 任务和子任务包括:
* Template filling: Extracting a fixed set of fields from a document, e.g. extract perpetrators, victims, time, etc. from a newspaper article about a terrorist attack.
** Event extraction: Given an input document, output zero or more event templates. For instance, a newspaper article might describe multiple terrorist attacks.
* [[Knowledge Base]] Population: Fill a database of facts given a set of documents. Typically the database is in the form of triplets, (entity 1, relation, entity 2), e.g. ([[Barack Obama]], Spouse, [[Michelle Obama]])
** [[Named entity recognition]]: recognition of known entity names (for people and organizations), place names, temporal expressions, and certain types of numerical expressions, by employing existing knowledge of the domain or information extracted from other sentences.<ref name="ecir2019">{{cite conference| author= Dat Quoc Nguyen and Karin Verspoor | title=End-to-end neural relation extraction using deep biaffine attention | book-title=Proceedings of the 41st European Conference on Information Retrieval (ECIR)| year=2019 |doi=10.1007/978-3-030-15712-8_47| arxiv=1812.11275}}</ref> Typically the recognition task involves assigning a unique identifier to the extracted entity. A simpler task is ''named entity detection'', which aims at detecting entities without having any existing knowledge about the entity instances. For example, in processing the sentence "M. Smith likes fishing", ''named entity detection'' would denote '''detecting''' that the phrase "M. Smith" does refer to a person, but without necessarily having (or using) any knowledge about a certain ''M. Smith'' who is (or, "might be") the specific person whom that sentence is talking about.
** [[Coreference]] resolution: detection of [[coreference]] and [[Anaphora (linguistics)|anaphoric]] links between text entities. In IE tasks, this is typically restricted to finding links between previously-extracted named entities. For example, "International Business Machines" and "IBM" refer to the same real-world entity. If we take the two sentences "M. Smith likes fishing. But he doesn't like biking", it would be beneficial to detect that "he" is referring to the previously detected person "M. Smith".
** [[Relationship extraction]]: identification of relations between entities,<ref name="ecir2019" /> such as:
*** PERSON works for ORGANIZATION (extracted from the sentence "Bill works for IBM.")
*** PERSON located in LOCATION (extracted from the sentence "Bill is in France.")
* Semi-structured information extraction which may refer to any IE that tries to restore some kind of information structure that has been lost through publication, such as:
** Table extraction: finding and extracting tables from documents.<ref>{{cite journal | vauthors = Milosevic N, Gregson C, Hernandez R, Nenadic G | title = A framework for information extraction from tables in biomedical literature | journal = International Journal on Document Analysis and Recognition (IJDAR) | volume = 22 | issue = 1 | pages = 55–78 | date = February 2019 | doi = 10.1007/s10032-019-00317-0 | arxiv = 1902.10031 | bibcode = 2019arXiv190210031M | s2cid = 62880746 }}</ref><ref>{{cite thesis |type=PhD |last=Milosevic |first=Nikola |date=2018 |title=A multi-layered approach to information extraction from tables in biomedical documents |publisher=University of Manchester | url=https://www.research.manchester.ac.uk/portal/files/70405100/FULL_TEXT.PDF}}</ref>
** Table information extraction : extracting information in structured manner from the tables. This is more complex task than table extraction, as table extraction is only the first step, while understanding the roles of the cells, rows, columns, linking the information inside the table and understanding the information presented in the table are additional tasks necessary for table information extraction. <ref>{{cite journal | vauthors = Milosevic N, Gregson C, Hernandez R, Nenadic G | title = A framework for information extraction from tables in biomedical literature | journal = International Journal on Document Analysis and Recognition (IJDAR) | volume = 22 | issue = 1 | pages = 55–78 | date = February 2019 | doi = 10.1007/s10032-019-00317-0 | arxiv = 1902.10031 | bibcode = 2019arXiv190210031M | s2cid = 62880746 }}</ref><ref>{{cite journal | vauthors = Milosevic N, Gregson C, Hernandez R, Nenadic G | title = Disentangling the structure of tables in scientific literature | journal = 21st International Conference on Applications of Natural Language to Information Systems | series = Lecture Notes in Computer Science | volume = 21 | date = June 2016 | pages = 162–174 | doi = 10.1007/978-3-319-41754-7_14 | isbn = 978-3-319-41753-0 | url = https://www.research.manchester.ac.uk/portal/en/publications/disentangling-the-structure-of-tables-in-scientific-literature(473111c2-52e9-493a-be8c-1a78c5b7ce36).html }}</ref><ref>{{cite thesis |type=PhD |last=Milosevic |first=Nikola |date=2018 |title=A multi-layered approach to information extraction from tables in biomedical documents |publisher=University of Manchester | url=https://www.research.manchester.ac.uk/portal/files/70405100/FULL_TEXT.PDF}}</ref>
** Comments extraction : extracting comments from actual content of article in order to restore the link between author of each sentence
* Language and vocabulary analysis
**[[Terminology extraction]]: finding the relevant terms for a given [[text corpus|corpus]]
Note that this list is not exhaustive and that the exact meaning of IE activities is not commonly accepted and that many approaches combine multiple sub-tasks of IE in order to achieve a wider goal. Machine learning, statistical analysis and/or natural language processing are often used in IE.
请注意，这一清单并非详尽无遗，而且普遍不接受 IE 活动的确切含义，许多方法将 IE 的多个子任务结合起来，以实现更广泛的目标。IE 中经常使用机器学习、统计分析和/或自然语言处理。
* Audio extraction
** Template-based music extraction: finding relevant characteristic in an audio signal taken from a given repertoire; for instance <ref>A.Zils, F.Pachet, O.Delerue and F. Gouyon, [http://www.csl.sony.fr/downloads/papers/2002/ZilsMusic.pdf Automatic Extraction of Drum Tracks from Polyphonic Music Signals], Proceedings of WedelMusic, Darmstadt, Germany, 2002.</ref> time indexes of occurrences of percussive sounds can be extracted in order to represent the essential rhythmic component of a music piece.
IE on non-text documents is becoming an increasingly interesting topic in research, and information extracted from multimedia documents can now be expressed in a high level structure as it is done on text. This naturally leads to the fusion of extracted information from multiple kinds of documents and sources.
非文本文档的 IE 正成为一个越来越引人注目的研究课题，从多媒体文档中提取的信息现在可以像在文本中一样以高层次的结构表达。这自然导致了从多种文档和资源中提取的信息的融合。
Note that this list is not exhaustive and that the exact meaning of IE activities is not commonly accepted and that many approaches combine multiple sub-tasks of IE in order to achieve a wider goal. Machine learning, statistical analysis and/or natural language processing are often used in IE.
IE has been the focus of the MUC conferences. The proliferation of the Web, however, intensified the need for developing IE systems that help people to cope with the enormous amount of data that is available online. Systems that perform IE from online text should meet the requirements of low cost, flexibility in development and easy adaptation to new domains. MUC systems fail to meet those criteria. Moreover, linguistic analysis performed for unstructured text does not exploit the HTML/XML tags and the layout formats that are available in online texts. As a result, less linguistically intensive approaches have been developed for IE on the Web using wrappers, which are sets of highly accurate rules that extract a particular page's content. Manually developing wrappers has proved to be a time-consuming task, requiring a high level of expertise. Machine learning techniques, either supervised or unsupervised, have been used to induce such rules automatically.
IE 已经成为 MUC 会议的焦点。然而，随着互联网的普及，人们更加需要开发 IE 系统，以帮助人们处理在线可用的大量数据。从在线文本执行 IE 的系统应该满足低成本、开发灵活性和易于适应新领域的要求。MUC 系统不能满足这些标准。此外，对非结构化文本执行的语言分析并没有利用 HTML/XML 标记和在线文本中可用的布局格式。因此，使用包装器为 IE 开发了语言密集度较低的方法，这些包装器是一组高度精确的规则，可以提取特定页面的内容。事实证明，手动开发包装器是一项耗时的任务，需要高水平的专业知识。机器学习技术，无论是监督或无监督，已被用来归纳这些规则自动。
IE on non-text documents is becoming an increasingly interesting topic{{when|date=March 2017}} in research, and information extracted from multimedia documents can now{{when|date=March 2017}} be expressed in a high level structure as it is done on text. This naturally leads to the fusion of extracted information from multiple kinds of documents and sources.
Wrappers typically handle highly structured collections of web pages, such as product catalogs and telephone directories. They fail, however, when the text type is less structured, which is also common on the Web. Recent effort on adaptive information extraction motivates the development of IE systems that can handle different types of text, from well-structured to almost free text -where common wrappers fail- including mixed types. Such systems can exploit shallow natural language knowledge and thus can be also applied to less structured texts.
Wrappers 通常处理高度结构化的网页集合，如产品目录和电话目录。然而，当文本类型结构化程度较低时，它们就会失败，这在 Web 上也很常见。最近在自适应信息抽取方面的努力促进了 IE 系统的发展，该系统可以处理不同类型的文本，从结构良好的到几乎是自由的文本——这是通常的包装器失败的地方——包括混合类型。这样的系统可以利用浅层的自然语言知识，因此也可以应用于结构化程度较低的文本。
==World Wide Web applications==
IE has been the focus of the MUC conferences. The proliferation of the [[World Wide Web|Web]], however, intensified the need for developing IE systems that help people to cope with the [[data deluge|enormous amount of data]] that is available online. Systems that perform IE from online text should meet the requirements of low cost, flexibility in development and easy adaptation to new domains. MUC systems fail to meet those criteria. Moreover, linguistic analysis performed for unstructured text does not exploit the HTML/[[XML]] tags and the layout formats that are available in online texts. As a result, less linguistically intensive approaches have been developed for IE on the Web using [[Wrapper (data mining)|wrappers]], which are sets of highly accurate rules that extract a particular page's content. Manually developing wrappers has proved to be a time-consuming task, requiring a high level of expertise. [[Machine learning]] techniques, either [[Supervised learning|supervised]] or [[Unsupervised learning|unsupervised]], have been used to induce such rules automatically.
A recent development is Visual Information Extraction, that relies on rendering a webpage in a browser and creating rules based on the proximity of regions in the rendered web page. This helps in extracting entities from complex web pages that may exhibit a visual pattern, but lack a discernible pattern in the HTML source code.
最近的一个发展是 Visual 信息抽取，它依赖于在浏览器中渲染网页，并根据渲染网页中区域的接近程度创建规则。这有助于从复杂的网页中提取实体，这些网页可能表现出一种视觉模式，但在 HTML 源代码中缺乏一种可识别的模式。
''Wrappers'' typically handle highly structured collections of web pages, such as product catalogs and telephone directories. They fail, however, when the text type is less structured, which is also common on the Web. Recent effort on ''adaptive information extraction'' motivates the development of IE systems that can handle different types of text, from well-structured to almost free text -where common wrappers fail- including mixed types. Such systems can exploit shallow natural language knowledge and thus can be also applied to less structured texts.
The following standard approaches are now widely accepted:
下列标准办法现已得到广泛接受:
A recent{{when|date=March 2017}} development is Visual Information Extraction,<ref>{{cite arXiv|eprint = 1506.08454|title=WYSIWYE: An Algebra for Expressing Spatial and Textual Rules for Information Extraction|first1=Vijil |last1=Chenthamarakshan|first2=Prasad M |last2=Desphande |first3= Raghu |last3=Krishnapuram |first4= Ramakrishnan |last4=Varadarajan |first5= Knut |last5=Stolze|year=2015|class=cs.CL}}</ref><ref>{{cite document|citeseerx = 10.1.1.21.8236|title=Visual Web Information Extraction with Lixto|first1=Robert |last1=Baumgartner|first2=Sergio |last2=Flesca |first3= Georg |last3=Gottlob|year=2001|pages=119–128}}</ref> that relies on rendering a webpage in a browser and creating rules based on the proximity of regions in the rendered web page. This helps in extracting entities from complex web pages that may exhibit a visual pattern, but lack a discernible pattern in the HTML source code.
==Approaches==
The following standard approaches are now widely accepted:
* Hand-written regular expressions (or nested group of regular expressions)
* Using classifiers
** Generative: [[naïve Bayes classifier]]
** Discriminative: [[Principle of maximum entropy#Maximum entropy models|maximum entropy models]] such as [[Multinomial logistic regression]]
* Sequence models
** [[Recurrent neural network]]
** [[Hidden Markov model]]
Numerous other approaches exist for IE including hybrid approaches that combine some of the standard approaches previously listed.
IE 还有许多其他方法，包括混合方法，它们结合了以前列出的一些标准方法。
** Conditional Markov model (CMM) / [[Maximum-entropy Markov model]] (MEMM)
** [[Conditional random field]]s (CRF) are commonly used in conjunction with IE for tasks as varied as extracting information from research papers<ref>{{Cite journal | doi = 10.1016/j.ipm.2005.09.002 | title = Information extraction from research papers using conditional random fields☆ | year = 2006 | last1 = Peng | first1 = F. | last2 = McCallum | first2 = A. | journal = Information Processing & Management | volume = 42 | issue = 4 | pages = 963}}</ref> to extracting navigation instructions.<ref>{{cite web|title=Extracting Frame-based Knowledge Representation from Route Instructions|last1=Shimizu|first1=Nobuyuki|last2=Hass|first2=Andrew|url=http://www.cs.albany.edu/~shimizu/shimizu+haas2006frame.pdf|year=2006|access-date=2010-03-27|archive-url=https://web.archive.org/web/20060901085639/http://www.cs.albany.edu/~shimizu/shimizu+haas2006frame.pdf|archive-date=2006-09-01|url-status=dead}}</ref>
Numerous other approaches exist for IE including hybrid approaches that combine some of the standard approaches previously listed.
==Free or open source software and services==
* [[General Architecture for Text Engineering]] (GATE) is bundled with a free Information Extraction system
* Apache [[OpenNLP]] is a Java machine learning toolkit for natural language processing
* [[ClearForest|OpenCalais]] is an automated information extraction web service from [[Thomson Reuters]] (Free limited version)
* [[Mallet (software project)|Machine Learning for Language Toolkit (Mallet)]] is a Java-based package for a variety of natural language processing tasks, including information extraction.
* [[DBpedia Spotlight]] is an open source tool in Java/Scala (and free web service) that can be used for named entity recognition and [[Name resolution (semantics and text extraction)|name resolution]].
* [[Natural Language Toolkit]] is a suite of libraries and programs for symbolic and statistical natural language processing (NLP) for the Python programming language
* See also [[Conditional random field#Software|CRF implementation]]s
==See also==
{{colbegin}}
* [[Ontology extraction]]
* [[Applications of artificial intelligence]]
* [[Concept mining]]
* [[DARPA TIPSTER Program]]
* [[Enterprise search]]
* [[Faceted search]]
* [[Knowledge extraction]]
* [[Named entity recognition]]
* [[Nutch]]
* [[Semantic translation]]
* [[Textmining]]
* [[Web scraping]]
Lists
名单
* [[Open information extraction]]
* [[Data extraction]]
; Lists
* [[List of emerging technologies]]
* [[Outline of artificial intelligence]]
{{colend}}
==References==
<references/>
{{refimprove|date=March 2017}}
== External links==
* [http://alias-i.com/lingpipe/web/competition.html Alias-I "competition" page] A listing of academic toolkits and industrial toolkits for natural language information extraction.
* [http://www.gabormelli.com/RKB/Information_Extraction_Task Gabor Melli's page on IE] Detailed description of the information extraction task.
Category:Natural language processing
类别: 自然语言处理
{{Natural Language Processing}}
Category:Artificial intelligence
类别: 人工智能
<noinclude>
<small>This page was moved from [[wikipedia:en:Information extraction]]. Its edit history can be viewed at [[信息抽取/edithistory]]</small></noinclude>
[[Category:待整理页面]]
{{short description|Automatically extracting structured information from un- or semi-structured machine-readable documents, such as human language texts}}
'''Information extraction''' ('''IE''') is the task of automatically extracting structured information from [[unstructured data|unstructured]] and/or semi-structured [[machine-readable data|machine-readable]] documents and other electronically represented sources. In most of the cases this activity concerns processing human language texts by means of [[natural language processing]] (NLP). Recent activities in [[multimedia]] document processing like automatic annotation and content extraction out of images/audio/video/documents could be seen as information extraction
Information extraction (IE) is the task of automatically extracting structured information from unstructured and/or semi-structured machine-readable documents and other electronically represented sources. In most of the cases this activity concerns processing human language texts by means of natural language processing (NLP). Recent activities in multimedia document processing like automatic annotation and content extraction out of images/audio/video/documents could be seen as information extraction
信息抽取是从非结构化和/或半结构化的机器可读文档和其他电子表示的源中自动提取结构化信息的任务。在大多数情况下，这种活动涉及到通过自然语言处理(NLP)来处理人类语言文本。最近在多媒体文档处理方面的活动，如图像/音频/视频/文档的自动注释和内容提取，可以被视为信息抽取
Due to the difficulty of the problem, current approaches to IE focus on narrowly restricted domains. An example is the extraction from newswire reports of corporate mergers, such as denoted by the formal relation:
Due to the difficulty of the problem, current approaches to IE focus on narrowly restricted domains. An example is the extraction from newswire reports of corporate mergers, such as denoted by the formal relation:
由于这个问题的难度，目前的 IE 方法集中在狭窄的限制领域。一个例子是从新闻通讯社关于公司合并的报告中摘录，例如用正式关系表示:
:<math>\mathrm{MergerBetween}(company_1, company_2, date)</math>,
<math>\mathrm{MergerBetween}(company_1, company_2, date)</math>,
公司1，公司2，日期,
from an online news sentence such as:
from an online news sentence such as:
从一个在线新闻句，如:
:''"Yesterday, New York based Foo Inc. announced their acquisition of Bar Corp."''
"Yesterday, New York based Foo Inc. announced their acquisition of Bar Corp."
昨天，总部位于纽约的 Foo 公司宣布收购 Bar corp.
A broad goal of IE is to allow computation to be done on the previously unstructured data. A more specific goal is to allow [[logical reasoning]] to draw inferences based on the logical content of the input data. Structured data is semantically well-defined data from a chosen target domain, interpreted with respect to category and [[context (language use)|context]].
A broad goal of IE is to allow computation to be done on the previously unstructured data. A more specific goal is to allow logical reasoning to draw inferences based on the logical content of the input data. Structured data is semantically well-defined data from a chosen target domain, interpreted with respect to category and context.
IE 的一个广泛的目标是允许在之前的非结构化数据上进行计算。一个更具体的目标是允许逻辑推理基于输入数据的逻辑内容做出推论。结构化数据是来自选定目标域的语义上定义良好的数据，根据类别和上下文进行解释。
Information Extraction is the part of a greater puzzle which deals with the problem of devising automatic methods for text management, beyond its transmission, storage and display. The discipline of [[information retrieval]] (IR)<ref>{{Cite journal|url = http://www.cs.bilkent.edu.tr/~guvenir/courses/CS550/Seminar/freitag2000-ml.pdf|title = Machine Learning for Information Extraction in Informal Domains|last = FREITAG|first = DAYNE|journal = 2000 Kluwer Academic Publishers. Printed in the Netherlands}}</ref> has developed automatic methods, typically of a statistical flavor, for indexing large document collections and classifying documents. Another complementary approach is that of [[natural language processing]] (NLP) which has solved the problem of modelling human language processing with considerable success when taking into account the magnitude of the task. In terms of both difficulty and emphasis, IE deals with tasks in between both IR and NLP. In terms of input, IE assumes the existence of a set of documents in which each document follows a template, i.e. describes one or more entities or events in a manner that is similar to those in other documents but differing in the details. An example, consider a group of newswire articles on Latin American terrorism with each article presumed to be based upon one or more terroristic acts. We also define for any given IE task a template, which is a(or a set of) case frame(s) to hold the information contained in a single document. For the terrorism example, a template would have slots corresponding to the perpetrator, victim, and weapon of the terroristic act, and the date on which the event happened. An IE system for this problem is required to “understand” an attack article only enough to find data corresponding to the slots in this template.
Information Extraction is the part of a greater puzzle which deals with the problem of devising automatic methods for text management, beyond its transmission, storage and display. The discipline of information retrieval (IR) has developed automatic methods, typically of a statistical flavor, for indexing large document collections and classifying documents. Another complementary approach is that of natural language processing (NLP) which has solved the problem of modelling human language processing with considerable success when taking into account the magnitude of the task. In terms of both difficulty and emphasis, IE deals with tasks in between both IR and NLP. In terms of input, IE assumes the existence of a set of documents in which each document follows a template, i.e. describes one or more entities or events in a manner that is similar to those in other documents but differing in the details. An example, consider a group of newswire articles on Latin American terrorism with each article presumed to be based upon one or more terroristic acts. We also define for any given IE task a template, which is a(or a set of) case frame(s) to hold the information contained in a single document. For the terrorism example, a template would have slots corresponding to the perpetrator, victim, and weapon of the terroristic act, and the date on which the event happened. An IE system for this problem is required to “understand” an attack article only enough to find data corresponding to the slots in this template.
信息抽取是一个更大的难题的一部分，它涉及的问题是设计文本管理的自动方法，超越了它的传输，存储和显示。信息检索学科已经开发出了自动化的方法，典型的统计方法，用于为大型文档集合建立索引和对文档进行分类。另一个互补的方法是自然语言处理(NLP) ，它解决了人类语言处理建模的问题，在考虑到任务的规模时取得了相当大的成功。就难度和重点而言，IE 处理介于 IR 和 NLP 之间的任务。在输入方面，IE 假定存在一组文档，其中每个文档都遵循一个模板，即。以类似于其他文档中的方式描述一个或多个实体或事件，但在细节上有所不同。例如，考虑一组关于拉丁美洲恐怖主义的新闻专线文章，每一条都被认为是基于一种或多种恐怖主义行为。我们还为任何给定的 IE 任务定义了一个模板，它是一个(或一组)案例框架，用于保存单个文档中包含的信息。对于恐怖主义的例子，一个模板应该有与恐怖主义行为的肇事者、受害者和武器相对应的位置，以及事件发生的日期。针对这个问题的 IE 系统需要“理解”一篇攻击文章，只要找到与此模板中插槽相对应的数据即可。
==History==
Information extraction dates back to the late 1970s in the early days of NLP.<ref>{{cite book|chapter-url=https://www.aclweb.org/anthology/A92-1024|chapter=Automatic Extraction of Facts from Press Releases to Generate News Stories|last1=Andersen|first1=Peggy M.|last2=Hayes|first2=Philip J.|citeseerx=10.1.1.14.7943|last3=Huettner|first3=Alison K.|last4=Schmandt|first4=Linda M.|last5=Nirenburg|first5=Irene B.|last6=Weinstein|first6=Steven P.|title=Proceedings of the third conference on Applied natural language processing -|year=1992|pages=170–177|doi=10.3115/974499.974531|s2cid=14746386}}</ref> An early commercial system from the mid-1980s was JASPER built for [[Reuters]] by the Carnegie Group Inc with the aim of providing [[Electronic communication network|real-time financial news]] to financial traders.<ref>{{cite book|url=http://pdfs.semanticscholar.org/2c90/fa59c6d9beed8dcb0e844725b872d3f33a35.pdf|archive-url=https://web.archive.org/web/20190220184608/http://pdfs.semanticscholar.org/2c90/fa59c6d9beed8dcb0e844725b872d3f33a35.pdf|url-status=dead|archive-date=2019-02-20|title=Information Extraction|last1=Cowie|first1=Jim|last2=Wilks|first2=Yorick|page=3|year=1996|citeseerx=10.1.1.61.6480|s2cid=10237124}}</ref>
Information extraction dates back to the late 1970s in the early days of NLP. An early commercial system from the mid-1980s was JASPER built for Reuters by the Carnegie Group Inc with the aim of providing real-time financial news to financial traders.
信息抽取可以追溯到20世纪70年代末 NLP 的早期。早期的商业系统是20世纪80年代中期由卡内基集团公司为路透社建立的 JASPER 系统，其目的是为金融交易员提供实时的财经新闻。
Beginning in 1987, IE was spurred by a series of [[Message Understanding Conference]]s. MUC is a competition-based conference<ref>Marco Costantino, Paolo Coletti, Information Extraction in Finance, Wit Press, 2008. {{ISBN|978-1-84564-146-7}}</ref> that focused on the following domains:
Beginning in 1987, IE was spurred by a series of Message Understanding Conferences. MUC is a competition-based conference that focused on the following domains:
从1987年开始，IE 受到了一系列信息理解会议的激励。MUC 是一个以竞争为基础的会议，重点关注以下领域:
*MUC-1 (1987), MUC-2 (1989): Naval operations messages.
*MUC-3 (1991), MUC-4 (1992): Terrorism in Latin American countries.
*MUC-5 (1993): [[Joint venture]]s and microelectronics domain.
*MUC-6 (1995): News articles on management changes.
*MUC-7 (1998): Satellite launch reports.
Considerable support came from the U.S. Defense Advanced Research Projects Agency ([[DARPA]]), who wished to automate mundane tasks performed by government analysts, such as scanning newspapers for possible links to terrorism.{{citation needed|date=March 2017}}
Considerable support came from the U.S. Defense Advanced Research Projects Agency (DARPA), who wished to automate mundane tasks performed by government analysts, such as scanning newspapers for possible links to terrorism.
美国国防部高级研究计划局(DARPA)提供了大量的支持，他们希望将政府分析人员执行的日常任务自动化，比如扫描报纸以寻找与恐怖主义的可能联系。
==Present significance==
The present significance of IE pertains to the growing amount of information available in unstructured form. [[Tim Berners-Lee]], inventor of the [[world wide web]], refers to the existing [[Internet]] as the web of ''documents'' <ref>{{cite web|url=http://tomheath.com/papers/bizer-heath-berners-lee-ijswis-linked-data.pdf|title=Linked Data - The Story So Far}}</ref> and advocates that more of the content be made available as a [[semantic web|web of ''data'']].<ref>{{cite web|url=http://www.ted.com/talks/tim_berners_lee_on_the_next_web.html|title=Tim Berners-Lee on the next Web}}</ref> Until this transpires, the web largely consists of unstructured documents lacking semantic [[metadata]]. Knowledge contained within these documents can be made more accessible for machine processing by means of transformation into [[relational database|relational form]], or by marking-up with [[XML]] tags. An intelligent agent monitoring a news data feed requires IE to transform unstructured data into something that can be reasoned with. A typical application of IE is to scan a set of documents written in a [[natural language]] and populate a database with the information extracted.<ref>[[Rohini Kesavan Srihari|R. K. Srihari]], W. Li, C. Niu and T. Cornell,"InfoXtract: A Customizable Intermediate Level Information Extraction Engine",[https://web.archive.org/web/20080507153920/http://journals.cambridge.org/action/displayIssue?iid=359643 Journal of Natural Language Engineering],{{dead link|date=September 2020}} Cambridge U. Press, 14(1), 2008, pp.33-69.</ref>
The present significance of IE pertains to the growing amount of information available in unstructured form. Tim Berners-Lee, inventor of the world wide web, refers to the existing Internet as the web of documents and advocates that more of the content be made available as a web of data. Until this transpires, the web largely consists of unstructured documents lacking semantic metadata. Knowledge contained within these documents can be made more accessible for machine processing by means of transformation into relational form, or by marking-up with XML tags. An intelligent agent monitoring a news data feed requires IE to transform unstructured data into something that can be reasoned with. A typical application of IE is to scan a set of documents written in a natural language and populate a database with the information extracted.
IE 目前的重要意义在于以非结构化的形式获得越来越多的信息。万维网的发明者 Tim Berners-Lee 将现有的互联网称为文档网络，并主张更多的内容以数据网络的形式提供。在此之前，网络大部分是由缺乏语义元数据的非结构化文档组成的。这些文档中包含的知识可以通过转换为关系形式或使用 XML 标记使机器处理更容易访问。一个监控新闻数据源的智能代理需要 IE 将非结构化数据变成可以理解的东西。IE 的典型应用程序是扫描一组用自然语言编写的文档，并用提取的信息填充数据库。
==Tasks and subtasks==
Applying information extraction to text is linked to the problem of [[text simplification]] in order to create a structured view of the information present in free text. The overall goal being to create a more easily machine-readable text to process the sentences. Typical IE tasks and subtasks include:
Applying information extraction to text is linked to the problem of text simplification in order to create a structured view of the information present in free text. The overall goal being to create a more easily machine-readable text to process the sentences. Typical IE tasks and subtasks include:
将信息抽取应用于文本是与文本简化问题联系在一起的，以便创建一个自由文本信息的结构化视图。总体目标是创建一个更容易机器阅读的文本来处理句子。典型的 IE 任务和子任务包括:
* Template filling: Extracting a fixed set of fields from a document, e.g. extract perpetrators, victims, time, etc. from a newspaper article about a terrorist attack.
** Event extraction: Given an input document, output zero or more event templates. For instance, a newspaper article might describe multiple terrorist attacks.
* [[Knowledge Base]] Population: Fill a database of facts given a set of documents. Typically the database is in the form of triplets, (entity 1, relation, entity 2), e.g. ([[Barack Obama]], Spouse, [[Michelle Obama]])
** [[Named entity recognition]]: recognition of known entity names (for people and organizations), place names, temporal expressions, and certain types of numerical expressions, by employing existing knowledge of the domain or information extracted from other sentences.<ref name="ecir2019">{{cite conference| author= Dat Quoc Nguyen and Karin Verspoor | title=End-to-end neural relation extraction using deep biaffine attention | book-title=Proceedings of the 41st European Conference on Information Retrieval (ECIR)| year=2019 |doi=10.1007/978-3-030-15712-8_47| arxiv=1812.11275}}</ref> Typically the recognition task involves assigning a unique identifier to the extracted entity. A simpler task is ''named entity detection'', which aims at detecting entities without having any existing knowledge about the entity instances. For example, in processing the sentence "M. Smith likes fishing", ''named entity detection'' would denote '''detecting''' that the phrase "M. Smith" does refer to a person, but without necessarily having (or using) any knowledge about a certain ''M. Smith'' who is (or, "might be") the specific person whom that sentence is talking about.
** [[Coreference]] resolution: detection of [[coreference]] and [[Anaphora (linguistics)|anaphoric]] links between text entities. In IE tasks, this is typically restricted to finding links between previously-extracted named entities. For example, "International Business Machines" and "IBM" refer to the same real-world entity. If we take the two sentences "M. Smith likes fishing. But he doesn't like biking", it would be beneficial to detect that "he" is referring to the previously detected person "M. Smith".
** [[Relationship extraction]]: identification of relations between entities,<ref name="ecir2019" /> such as:
*** PERSON works for ORGANIZATION (extracted from the sentence "Bill works for IBM.")
*** PERSON located in LOCATION (extracted from the sentence "Bill is in France.")
* Semi-structured information extraction which may refer to any IE that tries to restore some kind of information structure that has been lost through publication, such as:
** Table extraction: finding and extracting tables from documents.<ref>{{cite journal | vauthors = Milosevic N, Gregson C, Hernandez R, Nenadic G | title = A framework for information extraction from tables in biomedical literature | journal = International Journal on Document Analysis and Recognition (IJDAR) | volume = 22 | issue = 1 | pages = 55–78 | date = February 2019 | doi = 10.1007/s10032-019-00317-0 | arxiv = 1902.10031 | bibcode = 2019arXiv190210031M | s2cid = 62880746 }}</ref><ref>{{cite thesis |type=PhD |last=Milosevic |first=Nikola |date=2018 |title=A multi-layered approach to information extraction from tables in biomedical documents |publisher=University of Manchester | url=https://www.research.manchester.ac.uk/portal/files/70405100/FULL_TEXT.PDF}}</ref>
** Table information extraction : extracting information in structured manner from the tables. This is more complex task than table extraction, as table extraction is only the first step, while understanding the roles of the cells, rows, columns, linking the information inside the table and understanding the information presented in the table are additional tasks necessary for table information extraction. <ref>{{cite journal | vauthors = Milosevic N, Gregson C, Hernandez R, Nenadic G | title = A framework for information extraction from tables in biomedical literature | journal = International Journal on Document Analysis and Recognition (IJDAR) | volume = 22 | issue = 1 | pages = 55–78 | date = February 2019 | doi = 10.1007/s10032-019-00317-0 | arxiv = 1902.10031 | bibcode = 2019arXiv190210031M | s2cid = 62880746 }}</ref><ref>{{cite journal | vauthors = Milosevic N, Gregson C, Hernandez R, Nenadic G | title = Disentangling the structure of tables in scientific literature | journal = 21st International Conference on Applications of Natural Language to Information Systems | series = Lecture Notes in Computer Science | volume = 21 | date = June 2016 | pages = 162–174 | doi = 10.1007/978-3-319-41754-7_14 | isbn = 978-3-319-41753-0 | url = https://www.research.manchester.ac.uk/portal/en/publications/disentangling-the-structure-of-tables-in-scientific-literature(473111c2-52e9-493a-be8c-1a78c5b7ce36).html }}</ref><ref>{{cite thesis |type=PhD |last=Milosevic |first=Nikola |date=2018 |title=A multi-layered approach to information extraction from tables in biomedical documents |publisher=University of Manchester | url=https://www.research.manchester.ac.uk/portal/files/70405100/FULL_TEXT.PDF}}</ref>
** Comments extraction : extracting comments from actual content of article in order to restore the link between author of each sentence
* Language and vocabulary analysis
**[[Terminology extraction]]: finding the relevant terms for a given [[text corpus|corpus]]
Note that this list is not exhaustive and that the exact meaning of IE activities is not commonly accepted and that many approaches combine multiple sub-tasks of IE in order to achieve a wider goal. Machine learning, statistical analysis and/or natural language processing are often used in IE.
请注意，这一清单并非详尽无遗，而且普遍不接受 IE 活动的确切含义，许多方法将 IE 的多个子任务结合起来，以实现更广泛的目标。IE 中经常使用机器学习、统计分析和/或自然语言处理。
* Audio extraction
** Template-based music extraction: finding relevant characteristic in an audio signal taken from a given repertoire; for instance <ref>A.Zils, F.Pachet, O.Delerue and F. Gouyon, [http://www.csl.sony.fr/downloads/papers/2002/ZilsMusic.pdf Automatic Extraction of Drum Tracks from Polyphonic Music Signals], Proceedings of WedelMusic, Darmstadt, Germany, 2002.</ref> time indexes of occurrences of percussive sounds can be extracted in order to represent the essential rhythmic component of a music piece.
IE on non-text documents is becoming an increasingly interesting topic in research, and information extracted from multimedia documents can now be expressed in a high level structure as it is done on text. This naturally leads to the fusion of extracted information from multiple kinds of documents and sources.
非文本文档的 IE 正成为一个越来越引人注目的研究课题，从多媒体文档中提取的信息现在可以像在文本中一样以高层次的结构表达。这自然导致了从多种文档和资源中提取的信息的融合。
Note that this list is not exhaustive and that the exact meaning of IE activities is not commonly accepted and that many approaches combine multiple sub-tasks of IE in order to achieve a wider goal. Machine learning, statistical analysis and/or natural language processing are often used in IE.
IE has been the focus of the MUC conferences. The proliferation of the Web, however, intensified the need for developing IE systems that help people to cope with the enormous amount of data that is available online. Systems that perform IE from online text should meet the requirements of low cost, flexibility in development and easy adaptation to new domains. MUC systems fail to meet those criteria. Moreover, linguistic analysis performed for unstructured text does not exploit the HTML/XML tags and the layout formats that are available in online texts. As a result, less linguistically intensive approaches have been developed for IE on the Web using wrappers, which are sets of highly accurate rules that extract a particular page's content. Manually developing wrappers has proved to be a time-consuming task, requiring a high level of expertise. Machine learning techniques, either supervised or unsupervised, have been used to induce such rules automatically.
IE 已经成为 MUC 会议的焦点。然而，随着互联网的普及，人们更加需要开发 IE 系统，以帮助人们处理在线可用的大量数据。从在线文本执行 IE 的系统应该满足低成本、开发灵活性和易于适应新领域的要求。MUC 系统不能满足这些标准。此外，对非结构化文本执行的语言分析并没有利用 HTML/XML 标记和在线文本中可用的布局格式。因此，使用包装器为 IE 开发了语言密集度较低的方法，这些包装器是一组高度精确的规则，可以提取特定页面的内容。事实证明，手动开发包装器是一项耗时的任务，需要高水平的专业知识。机器学习技术，无论是监督或无监督，已被用来归纳这些规则自动。
IE on non-text documents is becoming an increasingly interesting topic{{when|date=March 2017}} in research, and information extracted from multimedia documents can now{{when|date=March 2017}} be expressed in a high level structure as it is done on text. This naturally leads to the fusion of extracted information from multiple kinds of documents and sources.
Wrappers typically handle highly structured collections of web pages, such as product catalogs and telephone directories. They fail, however, when the text type is less structured, which is also common on the Web. Recent effort on adaptive information extraction motivates the development of IE systems that can handle different types of text, from well-structured to almost free text -where common wrappers fail- including mixed types. Such systems can exploit shallow natural language knowledge and thus can be also applied to less structured texts.
Wrappers 通常处理高度结构化的网页集合，如产品目录和电话目录。然而，当文本类型结构化程度较低时，它们就会失败，这在 Web 上也很常见。最近在自适应信息抽取方面的努力促进了 IE 系统的发展，该系统可以处理不同类型的文本，从结构良好的到几乎是自由的文本——这是通常的包装器失败的地方——包括混合类型。这样的系统可以利用浅层的自然语言知识，因此也可以应用于结构化程度较低的文本。
==World Wide Web applications==
IE has been the focus of the MUC conferences. The proliferation of the [[World Wide Web|Web]], however, intensified the need for developing IE systems that help people to cope with the [[data deluge|enormous amount of data]] that is available online. Systems that perform IE from online text should meet the requirements of low cost, flexibility in development and easy adaptation to new domains. MUC systems fail to meet those criteria. Moreover, linguistic analysis performed for unstructured text does not exploit the HTML/[[XML]] tags and the layout formats that are available in online texts. As a result, less linguistically intensive approaches have been developed for IE on the Web using [[Wrapper (data mining)|wrappers]], which are sets of highly accurate rules that extract a particular page's content. Manually developing wrappers has proved to be a time-consuming task, requiring a high level of expertise. [[Machine learning]] techniques, either [[Supervised learning|supervised]] or [[Unsupervised learning|unsupervised]], have been used to induce such rules automatically.
A recent development is Visual Information Extraction, that relies on rendering a webpage in a browser and creating rules based on the proximity of regions in the rendered web page. This helps in extracting entities from complex web pages that may exhibit a visual pattern, but lack a discernible pattern in the HTML source code.
最近的一个发展是 Visual 信息抽取，它依赖于在浏览器中渲染网页，并根据渲染网页中区域的接近程度创建规则。这有助于从复杂的网页中提取实体，这些网页可能表现出一种视觉模式，但在 HTML 源代码中缺乏一种可识别的模式。
''Wrappers'' typically handle highly structured collections of web pages, such as product catalogs and telephone directories. They fail, however, when the text type is less structured, which is also common on the Web. Recent effort on ''adaptive information extraction'' motivates the development of IE systems that can handle different types of text, from well-structured to almost free text -where common wrappers fail- including mixed types. Such systems can exploit shallow natural language knowledge and thus can be also applied to less structured texts.
The following standard approaches are now widely accepted:
下列标准办法现已得到广泛接受:
A recent{{when|date=March 2017}} development is Visual Information Extraction,<ref>{{cite arXiv|eprint = 1506.08454|title=WYSIWYE: An Algebra for Expressing Spatial and Textual Rules for Information Extraction|first1=Vijil |last1=Chenthamarakshan|first2=Prasad M |last2=Desphande |first3= Raghu |last3=Krishnapuram |first4= Ramakrishnan |last4=Varadarajan |first5= Knut |last5=Stolze|year=2015|class=cs.CL}}</ref><ref>{{cite document|citeseerx = 10.1.1.21.8236|title=Visual Web Information Extraction with Lixto|first1=Robert |last1=Baumgartner|first2=Sergio |last2=Flesca |first3= Georg |last3=Gottlob|year=2001|pages=119–128}}</ref> that relies on rendering a webpage in a browser and creating rules based on the proximity of regions in the rendered web page. This helps in extracting entities from complex web pages that may exhibit a visual pattern, but lack a discernible pattern in the HTML source code.
==Approaches==
The following standard approaches are now widely accepted:
* Hand-written regular expressions (or nested group of regular expressions)
* Using classifiers
** Generative: [[naïve Bayes classifier]]
** Discriminative: [[Principle of maximum entropy#Maximum entropy models|maximum entropy models]] such as [[Multinomial logistic regression]]
* Sequence models
** [[Recurrent neural network]]
** [[Hidden Markov model]]
Numerous other approaches exist for IE including hybrid approaches that combine some of the standard approaches previously listed.
IE 还有许多其他方法，包括混合方法，它们结合了以前列出的一些标准方法。
** Conditional Markov model (CMM) / [[Maximum-entropy Markov model]] (MEMM)
** [[Conditional random field]]s (CRF) are commonly used in conjunction with IE for tasks as varied as extracting information from research papers<ref>{{Cite journal | doi = 10.1016/j.ipm.2005.09.002 | title = Information extraction from research papers using conditional random fields☆ | year = 2006 | last1 = Peng | first1 = F. | last2 = McCallum | first2 = A. | journal = Information Processing & Management | volume = 42 | issue = 4 | pages = 963}}</ref> to extracting navigation instructions.<ref>{{cite web|title=Extracting Frame-based Knowledge Representation from Route Instructions|last1=Shimizu|first1=Nobuyuki|last2=Hass|first2=Andrew|url=http://www.cs.albany.edu/~shimizu/shimizu+haas2006frame.pdf|year=2006|access-date=2010-03-27|archive-url=https://web.archive.org/web/20060901085639/http://www.cs.albany.edu/~shimizu/shimizu+haas2006frame.pdf|archive-date=2006-09-01|url-status=dead}}</ref>
Numerous other approaches exist for IE including hybrid approaches that combine some of the standard approaches previously listed.
==Free or open source software and services==
* [[General Architecture for Text Engineering]] (GATE) is bundled with a free Information Extraction system
* Apache [[OpenNLP]] is a Java machine learning toolkit for natural language processing
* [[ClearForest|OpenCalais]] is an automated information extraction web service from [[Thomson Reuters]] (Free limited version)
* [[Mallet (software project)|Machine Learning for Language Toolkit (Mallet)]] is a Java-based package for a variety of natural language processing tasks, including information extraction.
* [[DBpedia Spotlight]] is an open source tool in Java/Scala (and free web service) that can be used for named entity recognition and [[Name resolution (semantics and text extraction)|name resolution]].
* [[Natural Language Toolkit]] is a suite of libraries and programs for symbolic and statistical natural language processing (NLP) for the Python programming language
* See also [[Conditional random field#Software|CRF implementation]]s
==See also==
{{colbegin}}
* [[Ontology extraction]]
* [[Applications of artificial intelligence]]
* [[Concept mining]]
* [[DARPA TIPSTER Program]]
* [[Enterprise search]]
* [[Faceted search]]
* [[Knowledge extraction]]
* [[Named entity recognition]]
* [[Nutch]]
* [[Semantic translation]]
* [[Textmining]]
* [[Web scraping]]
Lists
名单
* [[Open information extraction]]
* [[Data extraction]]
; Lists
* [[List of emerging technologies]]
* [[Outline of artificial intelligence]]
{{colend}}
==References==
<references/>
{{refimprove|date=March 2017}}
== External links==
* [http://alias-i.com/lingpipe/web/competition.html Alias-I "competition" page] A listing of academic toolkits and industrial toolkits for natural language information extraction.
* [http://www.gabormelli.com/RKB/Information_Extraction_Task Gabor Melli's page on IE] Detailed description of the information extraction task.
Category:Natural language processing
类别: 自然语言处理
{{Natural Language Processing}}
Category:Artificial intelligence
类别: 人工智能
<noinclude>
<small>This page was moved from [[wikipedia:en:Information extraction]]. Its edit history can be viewed at [[信息抽取/edithistory]]</small></noinclude>
[[Category:待整理页面]]