CN101473304B - By the method that circuit simulation is graded to webpage - Google Patents
By the method that circuit simulation is graded to webpage Download PDFInfo
- Publication number
- CN101473304B CN101473304B CN200780022679.2A CN200780022679A CN101473304B CN 101473304 B CN101473304 B CN 101473304B CN 200780022679 A CN200780022679 A CN 200780022679A CN 101473304 B CN101473304 B CN 101473304B
- Authority
- CN
- China
- Prior art keywords
- node
- webpage
- resistance
- circuit model
- web
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Information Transfer Between Computers (AREA)
Abstract
Description
背景技术Background technique
【01】环球网(web)或因特网上的站点(网站)数量已经增长到数以十亿计。对于用户来说,用于在网上找到有用的站点和信息的有效方法变得越来越重要。包括GoogleTM、Ask.comTM,Yahoo!TM和MSNTM的互连网搜索引擎采用不同的方法来有效地搜索站点。站点搜索已经成为互连网用户日常生活的重要部分,其在带给人们信息方面具有重要的价值。[01] The number of sites (websites) on the World Wide Web (web) or the Internet has grown to billions. Efficient methods for finding useful sites and information on the web are becoming increasingly important to users. Including Google ™ , Ask.com ™ , Yahoo! The Internet search engines of TM and MSN TM use different methods to efficiently search sites. Site search has become an important part of Internet users' daily life, and it has great value in bringing people information.
【02】目前各种算法的站点搜索技术仍然存在许多缺点。Google技术公司在它的搜索引擎中采用PageRank(网页评级)和Hilltop方法。该方法根据站点间的超链接对站点评级。它采用“乘幂法”(powermethod)来查找所谓远传矩阵(teleportationmatrix)的特征向量。该特征向量给出站点的等级。为了确保会聚,采用等于0.85的远传参数。已经得到的报告是,当采用不同的远传参数时,PageRank方法会给出不同的结果。另一种算法的搜索方法(即Ask.comTM采用的ExpertRank(专家评级)方法)也解决了特征值和特征向量问题。该方法不仅仅识别具有与某一主题相关的信息的页面,其被称作“权威(authorities)”;还寻找那些包含许多到具有关于该主题的有用信息的页面的链接的页面,其称为“网络中心”。这是基于以下假设:关于给定主题的最权威的页面是那些最频繁地被其它页面所指向的页面。[02] At present, there are still many shortcomings in the site search technology of various algorithms. Google technology company adopts PageRank in its search engine (Web Page Rating) and the Hilltop method. This method ranks sites based on the hyperlinks between them. It uses a "powermethod" to find the eigenvectors of the so-called teleportation matrix. This eigenvector gives the rank of the site. To ensure convergence, a teleport parameter equal to 0.85 is used. It has been reported that the PageRank method gives different results when different remote parameters are used. Another algorithmic search method (ie, the ExpertRank (expert rating) method adopted by Ask.com TM ) also solves the problem of eigenvalues and eigenvectors. The method doesn't just identify pages with information on a topic, called "authorities"; it also looks for pages that contain many links to pages with useful information on the topic, called "authorities.""networkcentre". This is based on the assumption that the most authoritative pages on a given topic are those most frequently pointed to by other pages.
【03】现有算法易受垃圾链接的影响,其中垃圾站点通过创建大量指向它们的站点的链接来提高它们的级别。另外,一些现有算法不利于具有大量互连站点的大网域。[03] Existing algorithms are susceptible to link spam, where spammy sites boost their rank by creating a large number of links pointing to their sites. Additionally, some existing algorithms are not conducive to large network domains with a large number of interconnected sites.
【04】期望找到具有大量内容的站点和具有多个加权的入站链接的站点。而且还期望有利于网站管理员维护他们的网站并移除垃圾链接。[04] Expect to find sites with a lot of content and sites with multiple weighted inbound links. It is also expected to help webmasters maintain their sites and remove spam links.
发明内容Contents of the invention
【5】广义上讲,本发明通过提供一种用于经电路仿真对站点评定等级的方法来满足这些需要。[5] Broadly speaking, the present invention addresses these needs by providing a method for ranking sites via circuit simulation.
【6】应当理解到,本发明可以以多种方式实施,包括过程、设备、系统、装置或方法等。下面描述了本发明的多个有创造性的实施例。[6] It should be understood that the present invention can be implemented in various ways, including process, device, system, device or method and so on. A number of inventive embodiments of the invention are described below.
【7】在本发明的一个实施例中,提供了一种用于对站点评定等级的方法。所述方法包括生成具有表示每个站点的节点的网电路模型。仿真该模型以识别每个节点的电势。随后根据这些站点对应节点的电势可以对该站点评定等级。该网电路模型具有被配置为传输能量到该网电路的能量源。该能量源通过具有选定电阻的源连接器被连接到每个节点。另外,每个节点使用漏极连接被连接到能量漏极,该漏极连接具有与对应于节点的站点的内容分数以及入站链接的数量相关的电阻,还具有提供链接的相应网页的电压电平。对于连接诸站点中的第一站点到其中的第二站点的每个超链接,类二极管装置被添加到诸节点中的第一节点和第二节点之间的网电路中,第一节点对应于第一站点,第二节点对应于第二站点,该类二极管装置使电流从第一节点到第二节点的流动比从第二节点流到第一节点更容易。[7] In one embodiment of the present invention, a method for rating sites is provided. The method includes generating a network circuit model with nodes representing each site. Simulate the model to identify the potential at each node. The sites can then be graded according to the potentials of the corresponding nodes of these sites. The mesh circuit model has energy sources configured to transfer energy to the mesh circuit. The energy source is connected to each node through a source connector with a selected resistance. In addition, each node is connected to an energy drain using a drain connection that has a resistance related to the content score of the site corresponding to the node and the number of inbound links, as well as the voltage level of the corresponding web page providing the link. flat. For each hyperlink connecting a first of the sites to a second of the sites, a diode-like device is added to the network circuit between the first and second of the nodes, the first node corresponding to The first station, the second node corresponds to the second station, and the diode arrangement makes it easier for current to flow from the first node to the second node than from the second node to the first node.
【8】在另一个实施例中,提供了一种采用计算机系统执行网搜索(websearch)的方法。在该方法中,识别网络上的站点。对于每个站点,基于站点的内容给出内容分数。生成站点的索引,所述索引是通过从这些站点中找到的词语进行索引的计算机可搜索数据库。生成网电路模型,所述网电路模型包括表示每个站点的节点、能量源极、和对应于每个节点的源极连接端。每个源极连接端将能量源极连接到诸节点其中一个节点,每个漏极连接端将每个节点接地。所述源极连接端具有一选定电阻,该漏极连接端具有与该同节点相关联的站点的内容分数以及入站链接的数量相关的电阻和提供链接的相应网页的电压电平。连接两个站点的每个超链接在网电路中被表示为类二极管装置,该类二极管装置在对应于超链接方向的方向上具有低电阻,而在对应于超链接方向的相反方向上具有高电阻。采用计算机来仿真该网电路以确定每个节点的电势。根据相应节点的相应电压来对站点进行评级和分类。随后生成站点的子集列表,所述子集包括与查询相关的多个站点。[8] In another embodiment, a method of performing a web search (websearch) using a computer system is provided. In the method, a site on a network is identified. For each site, a content score is given based on the content of the site. An index of sites is generated, which is a computer-searchable database indexed by terms found in those sites. A grid circuit model is generated including nodes representing each station, energy sources, and source connections corresponding to each node. Each source connection connects the energy source to one of the nodes and each drain connection connects each node to ground. The source connection has a selected resistance and the drain connection has a resistance related to the content score of the site associated with the same node and the number of inbound links and the voltage level of the corresponding web page providing the link. Each hyperlink connecting two sites is represented in a web circuit as a diode-like device having a low resistance in the direction corresponding to the direction of the hyperlink and a high resistance in the opposite direction corresponding to the direction of the hyperlink. resistance. A computer is used to simulate the grid circuit to determine the potential of each node. Sites are rated and classified according to their respective voltages at their respective nodes. A list of a subset of sites is then generated, the subset comprising a plurality of sites relevant to the query.
【9】通过以下结合附图对本发明原理性的实例方式进行详细描述,本发明的优点将变得非常清楚。[9] The advantages of the present invention will become very clear through the following detailed description of the principle examples of the present invention in conjunction with the accompanying drawings.
附图说明Description of drawings
【10】通过以下结合附图对本发明的详细描述,本发明将可以得到充分理解。其中相同的参考标号代表相同的结构部件。[10] The present invention will be fully understood through the following detailed description of the present invention in conjunction with the accompanying drawings. Wherein the same reference numerals represent the same structural components.
【11】图1通过实例的方式,示出表示多个站点的网电路。[11] Figure 1 shows, by way of example, a network circuit representing multiple stations.
【12】图2通过实例的方式示出图1的网电路中两个节点间的电流与电势差关系的示意图。[12] Fig. 2 shows a schematic diagram of the relationship between current and potential difference between two nodes in the network circuit of Fig. 1 by way of example.
【13】图3A和3B通过实例的方式示出为深度分级站点进行增强评级的网电路变换。[13] Figures 3A and 3B illustrate, by way of example, network circuit transformations for enhanced ratings for deep rating sites.
【14】图4示出以实例方式表示用于利用被评级的站点执行网站搜索的程序流程图。[14] FIG. 4 shows, by way of example, a flow diagram of a program for performing a web site search using rated sites.
【15】图5示出以实例方式表示用于利用被评级在线百科全书网页发送广告的程序流程图。[15] FIG. 5 shows, by way of example, a flow diagram of a process for delivering advertisements using rated online encyclopedia pages.
【16】图6示出以实例方式表示用于集成网站目录和网站搜索以提高搜索相关性的过程的流程图。[16] FIG. 6 shows a flowchart representative by way of example of a process for integrating a website directory and website search to improve search relevancy.
具体实施方式detailed description
【17】在以下描述中,阐述了很多具体细节以提供对本发明的全面理解。但是,很显然对于本领域技术人员来说,在如果省略一个或多个具体细节的情况下,本发明也是可以实施的。在其它一些实例中,为了避免不必要地淡化本发明的特征,一些公知的处理操作和实现细节并未被详细描述。如此处所采用的,术语“网页”(“webpage”)表示具有单独网址的文档。术语“站点”(“website”)表示相关网页的集合,例如具有相同根网址的网上文档。短语“网”(“web”)表示通过因特网可访问的所有可索引的站点和网页。[17] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced without one or more of the specific details. In other instances, well known process operations and implementation details have not been described in detail in order not to unnecessarily obscure the aspects of the present invention. As used herein, the term "webpage" means a document having a single web address. The term "website" refers to a collection of related web pages, such as online documents, with the same root URL. The phrase "web" refers to all indexable sites and web pages accessible through the Internet.
【18】网建模【18】Network Modeling
【19】此处所描述的方法本质上是用于网页评级的非线性方法。经由“非线性”,意味着用于评级目的而描述网页的方程的系统是非线性的。通过在系统中非线性方程的采用,允许人们对网建模,以根据相关性更准确地对网页进行评级。在一个实施例中,网被模拟为集成电路,在此称之为“网电路”(“webcircuit”)。在网电路中,能够采用电路设计技术,来设计复杂的控制方案,用以对付垃圾站点。网电路模型也允许或者本地或者全局的对模型进行灵活的修改或调整以获取期望的结果。因为此处所描述的网电路模型是物理的,它总是具有物理解决方案。如下面将进一步详细描述的,此评级方法也能够考虑历史并确保评级变化中的平滑性。[19] The method described here is essentially a non-linear method for web page ranking. By "non-linear," it is meant that the system of equations describing web pages for rating purposes is non-linear. The adoption of non-linear equations in the system allows one to model the web to more accurately rank web pages in terms of relevance. In one embodiment, the web is modeled as an integrated circuit, referred to herein as a "web circuit." In the network circuit, circuit design technology can be used to design complex control schemes to deal with garbage sites. The network circuit model also allows for flexible modification or adjustment of the model, either locally or globally, to achieve desired results. Because the network circuit model described here is physical, it always has a physical solution. As will be described in further detail below, this rating method can also take into account history and ensure smoothness in rating changes.
【20】图1通过实例的方式示出表示多个网页的网电路100。每个网页被表示为网电路100中的一个节点102a-102d。网用户被模拟为一个电压节点104。电压节点104为网电路100提供能量源。在一个实施例中,电压节点104具有固定的电压,其电势可以被认为表示用户浏览网的意愿。在另一个实施例中,能量源具有固定电流。在不影响模型的可用性的情况下,可以设想使用其它的能量源。电阻器106a-106d连接用户与其想要访问的每个站点。如以下将要参考图4详细描述的,在一个实施例中,从由网索引的搜索中所获得的网页的列表中,检索表示为一种节点的每个网页。电阻器106a-106d的电阻可被认为用于代表用户访问该站点的困难度。电阻器106a-106d的电阻可以是固定的,亦即,对于一般搜索而言是一致的。对于依赖主题的搜索,如下面将更详细描述的,电阻可以是依赖主题的。例如:一个工程师用户能够与所有工程界或资源网页具有紧密连接,这可由到代表那些网页的节点的较低电阻来表示。因此,可通过改变用户和用户感兴趣的网页之间的电阻来设计复杂且个性化的模型。[20] FIG. 1 shows, by way of example, a web circuit 100 representing a plurality of web pages. Each web page is represented as a node 102a - 102d in the web circuit 100 . The grid subscriber is modeled as a voltage node 104 . Voltage node 104 provides an energy source for grid circuit 100 . In one embodiment, the voltage node 104 has a fixed voltage, and its potential can be considered to represent the user's willingness to browse the web. In another embodiment, the energy source has a fixed current. Other energy sources are conceivable without affecting the usability of the model. Resistors 106a-106d connect the user with each site he wants to visit. As will be described in detail below with reference to FIG. 4, in one embodiment, each web page represented as a type of node is retrieved from the list of web pages obtained in a search of the web index. The resistance of the resistors 106a-106d may be considered to represent the difficulty for the user to access the site. The resistance of the resistors 106a-106d may be fixed, that is, consistent for a general search. For topic-dependent searches, as will be described in more detail below, resistance may be topic-dependent. For example: an engineer user can have strong connections to all engineering community or resource web pages, which may be indicated by a lower resistance to nodes representing those web pages. Thus, complex and personalized models can be designed by varying the resistance between the user and the web pages the user is interested in.
【21】例如,在用户和可信任的索引目录站点(诸如Yahoo.com)之间可配置小电阻。因此,这些站点的电势非常接近于源极电势。索引目录站点一般是具有权威性的站点,其提供到与某些特定主题的信息相关的其它高质量站点的链接。典型地,用户非常了解这些站点且经常访问这些站点以找到其它站点。在实践中,网页的电势(例如:电压)可以是固定的,对于多个大索引可以是某个接近于源极电势的近似值。因此本质上,在仿真期间该节点能够被视为能量源极。该近似值能够通过减少电路的复杂性来节约仿真的成本。因为索引目录站点是权威性的,信任这些索引目录不会损害评级的质量。在一个实施例中,索引目录站点的质量是被主观确定的。[21] For example, a small resistor can be configured between the user and a trusted indexing directory site such as Yahoo.com. Therefore, the potential of these sites is very close to the source potential. Index directory sites are generally authoritative sites that provide links to other high-quality sites that have information on some particular topic. Typically, users are well aware of these sites and frequently visit these sites to find other sites. In practice, the potential (eg voltage) of the web page may be fixed, some approximation to the source potential for many large indexes. So essentially, the node can be seen as an energy source during the simulation. This approximation can save the cost of simulation by reducing the complexity of the circuit. Because indexed directory sites are authoritative, trusting these indexed directories will not compromise the quality of the ratings. In one embodiment, the quality of an indexed directory site is determined subjectively.
【22】在网电路模型中,线性电压可控电阻器108a-108d将每个节点102a-102d接地。线性电压可控电阻器108a-108d的电阻值可以与控制函数、出站超链接的总数量、内容分数(垃圾站点是低分数)以及从其它网页到该网页的加权入站链接成比例。加权入站链接只是利用基于多个不同因素配置给它的某个倍增因素的一种链接。例如,诸如从网页A到网页B的两个网页之间的超链接,可以通过在转化为电阻值中的几个因素而被加权;[22] In the mesh circuit model, a linear voltage controllable resistor 108a-108d connects each node 102a-102d to ground. The resistance values of the linear voltage controllable resistors 108a-108d may be proportional to the control function, the total number of outbound hyperlinks, the content score (spam is a low score), and the weighted inbound links to the webpage from other webpages. A weighted inbound link is simply a link that utilizes a certain multiplier factor assigned to it based on a number of different factors. For example, a hyperlink between two web pages, such as from web page A to web page B, can be weighted by several factors in converting to resistance value;
1.控制电压v(A,0)的线性函数,其中v(A,0)是节点A和地之间的电压差值;1. A linear function of the control voltage v(A, 0), where v(A, 0) is the voltage difference between node A and ground;
2.如果A不是索引目录站点的一部分网页,网页A的出站链接的总数的倒数;2. The inverse of the total number of outbound links to page A if A is not part of the indexed directory site;
3.如果A是索引目录站点,与索引目录站点的质量成比例的数值,其可以通过整个服务中的链接总数来进行测量,3. If A is an indexing directory site, a value proportional to the quality of the indexing directory site, which can be measured by the total number of links in the entire service,
4.网页A的域中链接每个网页的平均数值的倒数,4. The reciprocal of the average number of links per page in the domain of page A,
5.网页A的内容分数,和5. the content score of page A, and
6.与该链接相关的内容相关性分数(广告链接的内容相关性低)。这给出了基于超链接网页B的内容与给出该链接的网页A的内容的相关性的分数。在一个实施例中,通过搜索网页B中与网页A中相同的字词来确定该相关性。6. The content relevance score associated with the link (the ad link has low content relevance). This gives a score based on the relevancy of the content of the hyperlinked web page B to the content of the web page A that gave the link. In one embodiment, the relevance is determined by searching webpage B for the same terms as webpage A.
【23】内容分数可被用于降低基于内容的垃圾站点在网上和衰减网页上的影响力。所设计的网页仅为实现提高这些和某些相关页面的级别,而没有增进对于阅读者的效用,被称为“网上垃圾”(“webspam”)。衰减网页是指包含旧的和中断的超链接的网页,即,这些超链接不再指向现存的网页。中断的超链接表明网页很少被维护。[23] Content scores can be used to reduce the impact of content-based spam on the web and decay web pages. Web pages designed solely to enhance the status of these and certain related pages, without enhancing their usefulness to readers, are known as "webspam." Decayed web pages are web pages that contain old and broken hyperlinks, ie, these hyperlinks no longer point to existing web pages. Broken hyperlinks indicate that the webpage is rarely maintained.
【24】一个普通类型的网上垃圾通过从字典中抽取流行的词汇来被自动的生成。通过与那些普通的网页进行比较,一些启发式(heuristic)方法可被用于检查下述统计以识别它们:网页上的词汇数和词汇的平均长度、锚索文本的数量、可视内容的片段、网页的可压缩性、从全局公共词汇中抽取的页面片段、全局公共词汇的片段、独立的n元似然性(n-gramlikelihoods)、和条件n元似然性。已有的方法学可用于内容分析。例如,在此引用作为参考的文献包括:由AlexandrosNtoulas等人的论文《通过内容分析检测垃圾网页》("DetectingSpamWebPagesthroughContentAnalysis,")。垃圾网页和衰减网页(decayingwebpage)二者都具有低内容分。[24] A common type of spam is automatically generated by extracting popular words from a dictionary. Some heuristic methods can be used to examine the following statistics to identify them by comparing them with those of ordinary web pages: number of words on a web page and average length of words, amount of anchor text, fragments of visual content , the compressibility of web pages, page fragments extracted from the global common vocabulary, fragments of the global common vocabulary, independent n-gram likelihoods, and conditional n-gram likelihoods. Existing methodologies can be used for content analysis. For example, documents incorporated herein by reference include: "Detecting Spam Web Pages through Content Analysis," by Alexandros Ntoulas et al. Both spam and decaying web pages have low content scores.
【25】在实践中,上述第三项的数值被设置为接近1以用于诸如Yahoo等良好的索引目录站点,而如果A有多个出站的链接的话,上述第二项的数值则比1小很多。代表网页的节点和地之间的电导率值(其是电阻值的倒数),可被认为代表用户感觉无趣然后停止网上冲浪的可能性。因此,接地端可表示网上冲浪的终点。用于线性电压可控电阻器108a-108d的电阻值的相对量值是重要的。其通过对所有网页统一地考虑上述因素来精心计算。[25] In practice, the value of the third item above is set close to 1 for a good index directory site such as Yahoo, and if A has multiple outbound links, the value of the second item above is higher than 1 small lot. The conductivity value (which is the reciprocal of the resistance value) between a node representing a webpage and ground, can be considered to represent the likelihood that a user will get bored and stop surfing the web. Therefore, the ground terminal can represent the end point of web surfing. The relative magnitudes of the resistance values for the linear voltage controllable resistors 108a-108d are important. It is carefully calculated by taking the above factors into account uniformly for all pages.
【26】大量的加权入站链接和一个网页的出色内容分数被解释为该节点和地之间的大电阻值。大电阻值的小电导率反应了用户的能量或“电势的”趋势在节点处保持高的事实。另一方面,少量加权入站链接和网页的不足内容分数,被解释为节点和地之间的小电阻。小电阻值的高电导率使节点“接近”于地。在一个实施例中,如果具有太多的出站链接,则将通过降低电阻来阻碍非索引目录的站点。在实践中,可以通过人工手动检测网页是否是索引目录站点的一部分。[26] A large number of weighted inbound links and a web page's excellent content score are interpreted as a large resistance value between the node and ground. The small conductance at the large resistance value reflects the fact that the user's energy or "potential" tendency remains high at the node. On the other hand, lightly weighted inbound links and insufficient content scores of web pages, are interpreted as small resistances between nodes and ground. High conductivity with small resistance values keeps the node "close" to ground. In one embodiment, non-indexed directory sites will be blocked by lowering resistance if there are too many outbound links. In practice, it is possible to manually detect whether a web page is part of an indexed directory site by a human.
【27】为模拟诸如两个网页间的超链接的连接,在电路中,类二极管装置110a-110c被用于连接两个网页节点。类二极管装置110a-110c允许两个节点间的能量反馈和前馈,其与两节点间的电势差值成比例。集成电路中的真正的二极管具有允许前向电流并限制反向电流的性质。在一个实施例中,类二极管装置110a-110c是类似的。[27] To simulate a connection such as a hyperlink between two web pages, in a circuit, diode-like devices 110a-110c are used to connect two web page nodes. The diode-like devices 110a-110c allow energy feedback and feedforward between two nodes, which is proportional to the potential difference between the two nodes. A true diode in an integrated circuit has the property of allowing forward current and limiting reverse current. In one embodiment, the diode-like devices 110a-110c are similar.
【28】可以注意到,网电路100是无源的,仅无源装置被用于代表超链接。因此,网电路100是稳定的。相反,由Google技术公司使用的PageRank系统需要远传数值c,0<c<1(实践中是0.85),从而使其会聚。这样,在PageRank系统中,超链接将被模拟为电压可控电流源。[28] It can be noted that the web circuit 100 is passive and only passive devices are used to represent hyperlinks. Therefore, the mesh circuit 100 is stable. Instead, the PageRank used by Google Technologies The system needs to transmit the value c remotely, 0<c<1 (0.85 in practice), so as to make it converge. Thus, in PageRank In the system, the hyperlink will be modeled as a voltage-controllable current source.
【29】图2通过实例的方式示出了两个节点P和N间的电流相对电势差的图150。链接是从P到N。如果节点P的电势高于节点N的电势,该装置的电导率相对大。如果节点P的电势低于节点N的电势,该装置的电导率相对小。这确保具有较低级别的网页不会通过创建到其的链接影响高级别网页的级别。而且,如果具有高级别的网页A创建了到低级别的网页B的链接,具有低级别的网页B的级别将显著提高。另外,级别上接近的两个网页的级别将会变得更接近。应当注意到,B的加权入站链接的数量通过从A到B的链接而增长。但是,该影响仅与A的级别以及其它一些预定因素有关,而与A的级别和B的级别的差值无关。[29] Fig. 2 shows, by way of example, a graph 150 of current versus potential difference between two nodes P and N. Links are from P to N. If the potential of node P is higher than the potential of node N, the conductivity of the device is relatively large. If the potential of node P is lower than that of node N, the conductivity of the device is relatively small. This ensures that webpages with lower rankings do not affect the ranking of higher ranking webpages by creating links to them. Also, if a web page A with a high rank creates a link to a web page B with a low rank, the rank of the web page B with a low rank will increase significantly. In addition, the ranks of two pages that are close in rank will become closer. It should be noted that the number of weighted inbound links for B grows with links from A to B. However, this effect is only related to the grade of A and some other predetermined factors, and has nothing to do with the difference between the grades of A and B.
【30】在一个实施例中,网电路模型限制可能试图通过创建许多指向它们网页的链接来提高它们级别的垃圾网页。为实现此目标,可通过分析链接的内容来调谐类二极管装置110a-110c。例如,可以给相同页面上的大量链接以及广告链接以较小重要性。同样地,给带有相关内容的链接以较大重要性。[30] In one embodiment, the web circuit model restricts web spam that may attempt to increase their level by creating many links to their web pages. To achieve this goal, the diode-like devices 110a-110c may be tuned by analyzing the linked content. For example, a large number of links on the same page as well as advertising links can be given less importance. Likewise, give greater importance to links with related content.
【31】网电路100的类二极管装置以非线性方式限制反向和前向电流。如图2中所示,在一个实施例中,类二极管装置是4区分段的。此分段的线性对计算的效率是有用的。在I区,从节点N到节点P没有电导率且在从N到P方向的电流被固定于一个预设电流值。在II区,在从节点N到节点P的方向存在较小的电导率。在III区,在从节点P到节点N的方向提供较大的电导率。在IV区,在P和N之间提供相对大的电导率,仅比III区的略小。[31] The diode-like arrangement of the mesh circuit 100 limits reverse and forward currents in a non-linear manner. As shown in Figure 2, in one embodiment, the diode-like device is segmented in 4 zones. This piecewise linearity is useful for computational efficiency. In region I, there is no conductivity from node N to node P and the current in the direction from N to P is fixed at a preset current value. In region II, there is less conductivity in the direction from node N to node P. In region III, greater conductivity is provided in the direction from node P to node N. In region IV, a relatively large conductivity is provided between P and N, only slightly less than that of region III.
【32】既可本地地在网的部分中也可以全局地对这个复杂的非线性电导率模型进行调谐,以生成期望的仿真结果。例如,非线性类二极管装置110a-110c可被调谐以在网域内对付垃圾站点。特别的,对于相同网域内的超链接,所增加的级别分享是可以被允许的。这是通过提高表示单个域内网页的节点之间在分段线性区域I-IV区的电导率来实现。这对于内部超链接是合理的。[32] This complex nonlinear conductivity model can be tuned both locally in sections of the network and globally to generate the desired simulation results. For example, nonlinear diode-like devices 110a-110c may be tuned to combat spam within a network domain. In particular, increased levels of sharing are allowed for hyperlinks within the same domain. This is achieved by increasing the conductivity in the piecewise linear region I-IV between nodes representing web pages within a single domain. This is reasonable for internal hyperlinks.
【33】所设计的网模型的行为[33] Behavior of the designed net model
【34】网电路模型被配置为具有下述行为特征:[34] The network circuit model is configured to have the following behavioral characteristics:
1.如果网页具有高内容分数和大量加权入站链接,给该网页以高级别。因为网电路模型中的反向能量流受限,所以入站链接是来自低级别网页还是高级别网页是无关紧要的。1. If a web page has a high content score and a large number of weighted inbound links, give the web page a high rank. Because of the limited reverse energy flow in the web circuit model, it doesn't matter whether an inbound link comes from a low-level or high-level page.
2.如果网页具有高内容分数、多个入站链接和多个出站链接,该网页级别将被它链接指向的网页的级别所影响。如果它链接到低级别网页,将分享它自己的级别并稍微减低它的级别。此特征将给网络管理员以印象,使其适当地维护它的链接。2. If a web page has a high content score, multiple inbound links and multiple outbound links, the page rank will be affected by the rank of the web page it links to. If it links to a low-ranking page, it will share its own rating and downgrade it slightly. This feature will give the network administrator the impression to maintain its link properly.
3.给被识别为垃圾页面的网页以低内容分数。即使该网页具有多个出站链接,它自己的级别保持低且它不会影响其它网页的评级。3. Give low content scores to web pages identified as spam. Even if the web page has multiple outbound links, its own rank remains low and it does not affect the ratings of other web pages.
4.如果网页具有良好的内容分数但没有多个入站或出站链接,该网页可具有平均甚至高于平均的级别。4. If a webpage has a good content score but does not have multiple inbound or outbound links, the webpage can have an average or even above average rating.
5.如果网页具有多个中断链接,则降低该网页级别。5. If the webpage has multiple broken links, then downgrade the webpage.
6.与之前的方法不同,其阻碍具有大量内部链接的网页的大网域,本方法不具有这种不期望的性质。6. Unlike the previous method, which hinders large domains of web pages with a large number of internal links, the present method does not have this undesirable property.
【35】因此,该网电路模型能够找到具有大量内容的网页和具有多个加权入站链接的网页。另外,该网电路模型支持网站管理员维护他们的站点并移除垃圾链接。[35] Therefore, the web circuit model is able to find web pages with a large amount of content and web pages with multiple weighted inbound links. Additionally, the web circuit model supports webmasters in maintaining their sites and removing spam links.
【36】与仿真或评级的稳定性无关,基于在代表网页的一个节点和地之间配置电导率的灵活性,可采用分等级的网搜索方案。更具体地,可采用深度层级来分离一个站点的根网页。具有深层级的站点是指例如www.microsoft.com这样的站点,其具有源于一个根网页的多个页面,相对于这个根,以具有多个级别(例如,多于一个或两个级别)的层级而被编址。此类网页可被分离为两个拷贝:主拷贝和次拷贝。主拷贝继承该网页在网上的位置,并给出到次拷贝的链接(其仅链接到次拷贝)。根据关键字可以索引主拷贝,该关键字与根网页的整个层级(即根网页和所有子页面)极度相关。关键字可通过分析层级中的所有子页面来确定。例如,可以执行内容分析以识别子网页中的关键字。如果存在关键字的话,则可从次拷贝中移除这些关键字。[36] Regardless of simulation or rating stability, a hierarchical web search scheme can be employed based on the flexibility to configure the conductivity between a node representing a web and ground. More specifically, a depth hierarchy can be employed to separate a site's root web page. A site with deep hierarchies is one such as www.microsoft.com that has multiple pages originating from a root web page, relative to this root, to have multiple levels (e.g., more than one or two levels) level is addressed. Such web pages may be separated into two copies: a primary copy and a secondary copy. The primary copy inherits the web page's location on the web and gives a link to the secondary copy (it only links to the secondary copy). The master copy can be indexed according to keywords that are extremely relevant to the entire hierarchy of the root web page (ie, the root web page and all sub-pages). Keywords can be determined by analyzing all subpages in the hierarchy. For example, content analysis may be performed to identify keywords in sub-web pages. Keywords, if present, may be removed from the secondary copy.
【37】在仿真期间,对于每个拷贝创建各自的节点,如图3A和3B中以实例方式所示出。图3A以网电路的节点形式示出网的部分的初始表示120。为此说明目的,省略了能量源极和来自节点的能量漏极连接。每个箭头可表示从一个网页到另一个网页的超链接,或在网电路的上下文中的如上参考图1所述的连接两个节点的类二极管装置。具有深层级的站点122包括一个根网页124和多个其它网页。根网页124包括入站链接127和出站链接128。其它输入链接和输出链接(未示出)可连接到站点122中的其它网页。进而,根据一个实施例,如图3B的修改型表示120’所示,根网页124被分离为主拷贝125和次拷贝126。主拷贝125被连接到入站127和出站链接128,其在网上位于与网页124在初始表示120中类似的位置,使得代表主拷贝125的节点继承该初始位置。在所示的修改型表示120’中,一个单链接或类二极管装置130被从主拷贝125添加到次网页126中。[37] During simulation, a respective node is created for each copy, as shown by way of example in Figures 3A and 3B. Figure 3A shows an initial representation 120 of a portion of a mesh in the form of nodes of a mesh circuit. For purposes of illustration, the energy source and energy drain connections from the nodes are omitted. Each arrow may represent a hyperlink from one web page to another, or in the context of a web circuit a diode-like device connecting two nodes as described above with reference to FIG. 1 . A site 122 with a deep hierarchy includes a root web page 124 and multiple other web pages. Root web page 124 includes inbound links 127 and outbound links 128 . Other incoming and outgoing links (not shown) may connect to other web pages in site 122 . Furthermore, according to one embodiment, the root web page 124 is separated into a primary copy 125 and a secondary copy 126, as shown in the modified representation 120' of FIG. 3B. Master copy 125 is connected to inbound 127 and outbound links 128, which are located in a similar position on the web as web page 124 was in original representation 120, so that the node representing master copy 125 inherits the original position. In the modified representation 120' shown, a single link or diode-like device 130 has been added from the primary copy 125 to the secondary web page 126.
【38】使用此设计,当搜索相关关键字中的任何字时,可以容易地找到具有高评级的主拷贝125,由于多个入站链接127它将出现在搜索结果的顶端中。因为具有有用信息的子网页的低级别,以前的搜索方法难以在深层级中找到有用信息。另一方面,如果搜索关键字而不是相关联的关键字,将找到具有较低级别的次拷贝,但是因为级别较低,它将不是在搜索结果的顶端中。这种方案使得搜索具有深层级的站点更可行且更有用,并将倾向于引导用户到具有深层级的相关站点的根网页。包括相关子网页的搜索结果可被压缩至根网页的搜索结果条目中以增加有用性,允许用户点击按钮以展开并查看所有子网页。在一个实施例中,用户可以选择执行此类型的搜索,或基于如图3A所示的初始表示进行搜索。[38] Using this design, when searching for any word in a related keyword, one can easily find the master copy 125 with a high rating, which will appear in the top of the search results due to multiple inbound links 127. Because of the low level of sub-pages with useful information, previous search methods have difficulty finding useful information in deep levels. On the other hand, if you search for keywords other than the associated keywords, you will find a secondary copy with a lower rank, but because of the lower rank, it will not be in the top of the search results. This approach makes searching for sites with deep hierarchies more feasible and useful, and will tend to direct users to root web pages of related sites with deep hierarchies. Search results including related sub-pages can be compressed into the search result entry of the root page for added usefulness, allowing the user to click a button to expand and view all sub-pages. In one embodiment, the user may choose to perform this type of search, or conduct a search based on the initial representation as shown in Figure 3A.
【39】实例【39】Example
【40】表1表示用于描述具有15个节点的网电路的示意性网表:[40] Table 1 represents a schematic netlist for describing a net circuit with 15 nodes:
【41】表2示出解决用于每个节点的电压的网表之后的结果:[41] Table 2 shows the results after solving the netlist for the voltage of each node:
【42】如表2所示,节点5具有最高分数,4.47318,因为相应的网页具有多个入站链接。节点2、3、4、5、6、7、8、9和13具有相对低的分数,因为它们没有入站链接和只拥有平庸的内容分数。这对于它们链接到节点5没有帮助。因为接地的电阻的调谐,节点11具有第二高的电压。节点10具有第三高的分数,是因为它的内容良好且其具有来自节点5的入站链接。节点12具有第四的分数,是因为它具有来自节点2的入站链接。因为节点2的电压低于节点5的电压,所以节点12的电压低于节点10的电压。在用户和节点13表示的网页之间不存在链接。因此,节点13具有最低的级别。这是一种个人化的网搜索并且将在后面加以解释。[42] As shown in Table 2, node 5 has the highest score, 4.47318, because the corresponding web page has multiple inbound links. Nodes 2, 3, 4, 5, 6, 7, 8, 9 and 13 have relatively low scores because they have no inbound links and only mediocre content scores. It doesn't help that they link to node 5. Node 11 has the second highest voltage because of the tuning of the resistance to ground. Node 10 has the third highest score because its content is good and it has an inbound link from Node 5. Node 12 has the fourth score because it has an inbound link from Node 2. Because the voltage at node 2 is lower than the voltage at node 5 , the voltage at node 12 is lower than the voltage at node 10 . There is no link between the user and the web page represented by node 13 . Therefore, node 13 has the lowest rank. This is a personalized web search and will be explained later.
【43】此处描述的非线性网电路模型具有较大的用于复杂的网评级系统设计的空间。用于分析集成电路(很多集成电路具有数百万节点)的现有技术可被用来采用本发明的网电路模型进行页面分级。为有效解决大的非线性系统,在此引入2006年12月18日提交的共同未决美国第11/612,335号专利申请作为参考。[43] The nonlinear network circuit model described here has a large space for the design of complex network rating systems. Existing techniques for analyzing integrated circuits (many with millions of nodes) can be used for page ranking using the web circuit model of the present invention. To efficiently address large nonlinear systems, co-pending US Patent Application No. 11/612,335, filed December 18, 2006, is hereby incorporated by reference.
【44】集成的网索引目录和网搜索[44] Integrated web index and web search
【45】与仿真或评级的稳定性无关,基于处理索引目录的灵活性,可提供集成的网搜索和索引目录服务。在一个实施例中,人工创建索引目录以引导网搜索,同时网搜索结果被用于帮助扩展索引目录。在索引目录和网搜索之间的正反馈能帮助其互相提高。因此,无需对网搜索用户接口做任何改变或依赖用户经验,网索引目录和网搜索之间的集成就可被自然地构建到该搜索过程中。[45] Regardless of simulation or rating stability, integrated web search and indexing services can be provided based on flexibility in handling indexing lists. In one embodiment, an index is manually created to guide web searches, while web search results are used to help expand the index. Positive feedback between indexed directories and web search can help each improve the other. Thus, without any changes to the web search user interface or reliance on user experience, integration between web indexes and web searches can be built into the search process naturally.
【46】在一个实施例中,人工创建并模拟初始的索引目录作为部分因特网空间的网电路。在网电路模型中,提升出站链接的权重。当模拟该网电路模型时,出站链接的权重会导致对搜索结果具有重大影响的网索引目录。用户仍然执行相同类型的搜索,但由于引导我们的后台评级的网索引目录得到改进和调整,搜索结果将变得更加相关。[46] In one embodiment, an initial index directory is manually created and simulated as a network circuit for a portion of the Internet space. In the network circuit model, the weight of outbound links is increased. When simulating this web circuit model, the weight of outbound links leads to web indexes that have a significant impact on search results. Users still perform the same types of searches, but the results will become more relevant thanks to improvements and tweaks to the web index that guides our backend ratings.
【47】图6示出表示将索引目录集成到网搜索之中的示意性过程的流程图300。如开始框302所示该过程起始,并流到创建初始网索引目录的操作304。可以人工地生成初始网索引目录或者可以使用现有的网索引目录。例如,初始网索引目录可以包括提供与金融、新闻、旅行等相关的站点的链接的大量网页。[47] FIG. 6 shows a flowchart 300 representing an exemplary process for integrating indexed directories into web search. The process begins as indicated by start block 302 and flows to operation 304 of creating an initial web index directory. The initial web index can be manually generated or an existing web index can be used. For example, an initial web index may include a number of web pages providing links to sites related to finance, news, travel, and the like.
【48】一旦在操作304中创建初始网索引目录,过程将流到操作306,其中,索引目录中的链接在评级过程中被给予更大的权重。这与现有网评级算法不同;在现有算法中采用的是每个向外链接的权重除以出站链接的总数。由于不执行这种除法运算,该链接被授予大的权重以在结果评级中提供较大提升。[48] Once the initial web index is created in operation 304, the process will flow to operation 306, where links in the index are given greater weight in the ranking process. This is different from the existing web rating algorithm; in the existing algorithm, the weight of each outgoing link is divided by the total number of outgoing links. Since this division operation is not performed, the link is awarded a large weight to provide a large boost in the resulting rating.
【49】随后在操作308中,如上参考图1的所述的构建一个网电路模型,包括网索引目录,并使其链接加权以增加它们的级别。在操作310中,仿真该网电路模型,并根据如上所述的仿真结果评级该网页。[49] Then in operation 308, a network circuit model is constructed as described above with reference to FIG. 1, including the network index, and its links are weighted to increase their ranks. In operation 310, the web circuit model is simulated, and the web page is rated according to the simulation results as described above.
【50】在操作312,仿真的结果被用于改进网索引目录,其依次将提升用于用户的搜索结果。在一个实施例中,操作312是一个手动和自动结合的过程。例如,不包括在索引目录中的顶端搜索结果可被验证且如果获得授权将被添加到索引目录。这进一步增强网索引目录以用于未来的搜索。[50] At operation 312, the results of the simulation are used to improve the web index, which in turn will improve the search results for the user. In one embodiment, operation 312 is a combined manual and automated process. For example, top search results not included in the index may be validated and added to the index if authorized. This further enhances the web index for future searches.
【51】在操作312后,该过程返回到操作306。因此,操作306-312以迭代方式可被重复任意次以持续改进集成的索引目录和搜索结果。[51] After operation 312, the process returns to operation 306. Accordingly, operations 306-312 may be repeated any number of times in an iterative manner to continuously improve the integrated index catalog and search results.
【52】查询敏感(Query-Sensitive)网搜索[52] Query-Sensitive Web Search
【53】网电路模型允许查询敏感网评级。这对于企业搜索尤其有用。术语“企业搜索”指在企业内部互联网或企业网站域名内的搜索。企业搜索内的网页数量比因特网上的要小得多。对于用户查询的相关网页的数量甚至更加有限。结果是,在每次查询后可以计算实时的网页评级。[53] The network circuit model allows querying of sensitive network ratings. This is especially useful for enterprise searches. The term "corporate search" refers to searches within a corporate intranet or corporate website domain name. The number of web pages within Enterprise Search is much smaller than on the Internet. The number of relevant web pages for a user query is even more limited. As a result, real-time web page rankings can be calculated after each query.
【54】在一个实施例中,可以增加在电压节点104和表示与查询无关的网页的节点之间的电阻器106a-106d的电阻。另外,可以减小在地和表示与查询无关的网页的节点之间的电阻器110a-110d的电阻。增加和/或减小的量可根据它们相关或不相关的程度进行调整。可采用关键字分析方法和其它公知的文档分析方法来确定相关度。因此这将降低不相关网页的级别。[54] In one embodiment, the resistance of resistors 106a-106d between voltage node 104 and nodes representing web pages not relevant to the query may be increased. Additionally, the resistance of resistors 110a-110d between ground and nodes representing web pages not relevant to the query can be reduced. The amount of increase and/or decrease can be adjusted according to the degree to which they are related or irrelevant. Relevance can be determined using keyword analysis methods and other known document analysis methods. So this will downgrade irrelevant pages.
【55】可选地,可通过修改内容分数将每个网页与查询的相关度加入到模型中。与查询更相关的网页可使它们的内容分数陡增,同时与查询相关度较小的网页使它们的内容分数陡降。此改变的结果将基于相关度修改代表网页的每个节点和地之间的电阻值。对于高相关度的网页,通过增大该节点和地之间的电阻值,可提高该网页的级别。[55] Optionally, the relevance of each webpage to the query can be incorporated into the model by modifying the content score. Pages that are more relevant to the query can have their Content Score spiked, while pages that are less relevant to the query can have their Content Score dropped. The result of this change will be to modify the resistance value between each node representing the webpage and ground based on the degree of correlation. For webpages with high correlation, the level of the webpage can be improved by increasing the resistance value between the node and the ground.
【56】对于查询敏感网页评级,具有来自其它与查询相关网页的多个入站链接的查询相关网页将被给予高评级。对于企业查询,可允许用户或企业控制这种调整。这将允许各种企业基于此处所述的网电路模型,利用搜索引擎获得他们各自所需的结果。[56] For query-sensitive webpage ratings, query-related webpages with multiple inbound links from other query-related webpages will be given high ratings. For enterprise queries, the user or enterprise may be allowed to control this adjustment. This will allow businesses of all kinds to use search engines to get the results they want based on the network circuit model described here.
【57】为了提高查询敏感网电路仿真中的效率和响应时间,可采用2006年12月18日提交的题目为《parallelMulti-RateCircuitSimulation(并联多率电路仿真)》的美国第11/612,335号专利申请中所述的多速率电路仿真方案。不相关的网页快速收敛为低分数并变为隐藏的。这些隐藏网页然后得以从仿真之中排除,从而简化该方案。[57] In order to improve the efficiency and response time in querying sensitive network circuit simulation, the US patent application No. 11/612,335 entitled "parallelMulti-Rate Circuit Simulation (parallel multi-rate circuit simulation)" submitted on December 18, 2006 can be used The multirate circuit emulation scheme described in . Irrelevant web pages quickly converge to low scores and become hidden. These hidden web pages are then excluded from the simulation, thereby simplifying the scheme.
【58】在非企业环境中,例如,在全球因特网搜索中,预设数量的顶端预评级网页可被模拟到一个网电路之中,以定制顶端预评级站点的级别。例如,可以根据查询,对从普通索引搜索返回的前1000000个网页自定义建模,以生成自定义的基于查询的评级。[58] In a non-corporate environment, eg, in a global Internet search, a preset number of top pre-rated web pages can be modeled into a web circuit to customize the ranking of the top pre-rated sites. For example, the top 1,000,000 web pages returned from a general index search can be custom modeled according to the query to generate a custom query-based rating.
【59】主题敏感网搜索【59】Subject Sensitive Web Search
【60】网电路模型也允许主题敏感的网搜索。在主题敏感网搜索中,识别与用户感兴趣的特定主题相关的索引目录服务。给这些索引目录服务的出站链接以较高的权重。在一个实施例中,调谐用于这些索引目录站点和索引目录内的站点之间的超链接的类二极管非线性电阻器模型,以允许更高的反向电导率。这将对在仿真中自动拣出良好的索引目录服务起作用。为实现此目的,也可根据网索引目录中的出站链接的数量标准化模型中的电导率。基于该设置,网电路仿真将拣出良好的特定主题索引目录以及良好的特定主题站点。[60] The web circuit model also allows topic-sensitive web searches. In topic-sensitive web search, an indexing directory service is identified that is relevant to a particular topic of interest to a user. Give higher weight to outbound links from these indexed directory services. In one embodiment, the diode-like varistor model for these indexed directory sites and hyperlinks between sites within the indexed directory is tuned to allow for higher reverse conductivity. This will work to automatically pick out good indexing directory services in the simulation. Conductivity in the model can also be normalized by the number of outbound links in the web index for this purpose. Based on this setting, Web Circuit Emulation will pick out good subject-specific index directories as well as good subject-specific sites.
【61】对于许多预先的标准主题,可执行主题敏感网搜索。可根据用户配置文件计算特定主题评级的加权平均,其可在查询时间完成。用户配置文件由对应于不同主题的不同兴趣程度来定义,而用于平均的加权则与兴趣程度成比例。不同主题中的用户兴趣程度可由用户直接设定或从用户过去网搜索的历史计算出来。[61] For a number of pre-standard topics, a topic-sensitive web search can be performed. A weighted average of topic-specific ratings can be calculated based on user profiles, which can be done at query time. User profiles are defined by different levels of interest corresponding to different topics, and the weighting used for averaging is proportional to the level of interest. The degree of user interest in different topics can be set directly by the user or calculated from the user's past web search history.
【62】交互式和自适应网搜索[62] Interactive and Adaptive Web Search
【63】在交互式的网评级模型中,用户可查询某些关键字并获得第一份网搜索结果。然后,允许用户从顶级结果中去除一些结果并请求另一次搜索。在下一个搜索中,执行修改的网电路仿真。在修改型网电路中,在被用户去除的网页和地之间设置电阻值非常低的电阻器。此修改型网电路仿真将更准确并且更具可能回到用户真正所需的结果。[63] In the interactive web rating model, users can query certain keywords and get the first web search results. Then, allow the user to drop some results from the top results and request another search. In the next search, a modified network circuit simulation is performed. In modified mesh circuits, a very low value resistor is placed between the web removed by the user and ground. This modified network circuit simulation will be more accurate and more likely to return the results that the user really wants.
【64】为了提高用于修改型网电路仿真的效率和响应时间,精选步骤可被限制在前1000000个条目。可以假定,在普通用户环境中,是没有人会浏览超过1000000个条目的。当然,根据应用可以选择其它数目的条目。[64] To improve efficiency and response time for modified network circuit simulations, the refinement step can be limited to the first 1,000,000 entries. It can be assumed that in an average user environment, no one will browse more than 1,000,000 entries. Of course, other numbers of entries may be selected depending on the application.
【65】另外,在实践应用中可采用网电路模型进行统计。在搜索之后具有多次点击的网页可被提升。该提升可通过增加网页节点和地之间的电阻来容易地完成。[65] In addition, the network circuit model can be used for statistics in practical applications. Web pages with multiple hits after a search can be promoted. This boost can easily be done by increasing the resistance between the web node and ground.
【66】由于因特网上的站点数量和信息总量激增,交互式和自适应网搜索将更有可能找到有用的结果。[66] As the number of sites and the amount of information on the Internet proliferate, interactive and adaptive web searches will become more likely to find useful results.
【67】评级变化中的平滑[67] Smoothing in rating changes
【68】特定网页的评级中的突发变化可能是不希望的。这样的变化可由去往或来自网页的链接的变化而引起,或由某一网虫的多次访问之间网页内容的变化所导致。采用此处描述的复杂网电路仿真方法,可以使突发的级别变化问题最小化。特别地,可使用动态模型以使级别变化更平滑。在动态评级模型中,每个节点都连接接地电容且及时执行网电路的瞬变过程仿真。接地的电容为节点电势的变化带来平滑性。图1中的类二极管装置110a-110c也可被分配以一定的电容值。[68] Sudden changes in the ratings of particular web pages may be undesirable. Such changes may be caused by changes in links to or from the web page, or by changes in the content of the web page between visits by a worm. Using the complex network circuit simulation approach described here, the problem of sudden level changes can be minimized. In particular, a dynamic model can be used to make level changes smoother. In the dynamic rating model, each node is connected to ground capacitance and a transient process simulation of the grid circuit is performed in time. Capacitance to ground brings smoothness to changes in node potential. The diode-like devices 110a-110c in FIG. 1 may also be assigned a certain capacitance value.
【69】另外,渐变也可应用到新站点。新站点和网上其余站点之间的连接可通过改变类二极管装置中的电导率及其接地电阻来控制。可在类二极管装置中使用小电导,在接地电阻中使用大电导,其限制新站点的评级,并因此限制对其它站点级别产生影响。一段时间之后,电导率和电阻可爬升到正常值。[69] Alternatively, gradients can also be applied to new sites. The connection between the new site and the remaining sites on the network can be controlled by varying the conductivity in the diode-like device and its resistance to ground. Small conductances can be used in diode-like devices and large conductances in ground resistances, which limit the rating of new sites and thus the impact on other site levels. After a period of time, the conductivity and resistance can climb to normal values.
【70】硬件仿真加速【70】Hardware simulation acceleration
【71】网电路仿真模型可由硬件加速器辅助。现场可编程门阵列(FPGA)可被用于设计诸如稀疏矩阵求解器的专用系统。该专用FPGA系统或专用集成电路芯片将提高电路仿真的速度。[71] Network circuit simulation models can be assisted by hardware accelerators. Field Programmable Gate Arrays (FPGAs) can be used to design specialized systems such as sparse matrix solvers. The dedicated FPGA system or ASIC chip will increase the speed of circuit simulation.
【72】网搜索工具的示意性应用[72] Schematic application of web search tool
【73】图4示出了以实例方式示出用已评级的网页执行网搜索的过程的流程图200。如开始框202所示该过程起始,进而执行操作204。在此操作中,因特网被初始地(或连续地)访问以识别网页以及网页间的超链接。然后,在操作206中,采用内容分析识别初步内容分数。更具体地,在操作204中识别的每个网页根据其内容被给定一个初步内容分数。初步内容分数可以如上参考图1所述地加以确定。[73] FIG. 4 shows a flowchart 200 illustrating by way of example a process for performing a web search with rated web pages. The process begins as indicated by start block 202 and proceeds to operation 204 . In this operation, the Internet is initially (or continuously) accessed to identify web pages and hyperlinks between web pages. Then, in operation 206, a preliminary content score is identified using content analysis. More specifically, each web page identified in operation 204 is given a preliminary content score based on its content. Preliminary content scores may be determined as described above with reference to FIG. 1 .
【74】在为操作204中所识别的网页建立初步内容分数后,该过程执行操作208,在此生成索引。该索引包括由网页中发现的词语索引的网页地址的数据库。另外,该索引可包括每个网页的初步内容分数以及其它信息,例如出站超链接和来自其它网页的加权入站超链接的总数,如上参考图1所述。在一个实施例中,可基于因特网作为没有首先接收到查询的整体来建立评级。[74] After establishing preliminary content scores for the web pages identified in operation 204, the process proceeds to operation 208, where an index is generated. The index includes a database of web page addresses indexed by terms found in web pages. Additionally, the index may include a preliminary content score for each web page, as well as other information, such as the total number of outbound hyperlinks and weighted inbound hyperlinks from other web pages, as described above with reference to FIG. 1 . In one embodiment, the rating may be established based on the Internet as a whole without first receiving the query.
【75】根据评级是否为查询敏感模式可选地执行操作210。在查询敏感的情况下,在操作212建立网电路之前获得查询。为获得查询,用户输入查询字符串,例如一个或多个与用户正在查找的信息相关的字词或短语。当以网页的方式被实现时,用户将查询输入到网页浏览器窗口中的文本框。查询字符串与索引中的网页进行匹配以获得与查询字符串相关的网页的列表。[75] Operation 210 is optionally performed according to whether the rating is query sensitive mode. Where the query is sensitive, the query is obtained prior to establishing the network circuit at operation 212 . To obtain a query, a user enters a query string, such as one or more words or phrases related to the information the user is looking for. When implemented as a web page, a user enters a query into a text box in a web browser window. The query string is matched against web pages in the index to obtain a list of web pages related to the query string.
【76】在操作212,建立网电路。如上参考图1所述的电路那样,该网电路模拟该索引,或者该索引的子集。如果评级是查询敏感的,如上所述,根据通过搜索检索的每个网页与该查询的相关度来修改该电路。在一个实施例中,根据通过先执行操作212-218的查询敏感的评级来对网页进行预评级,如下面将进行详细描述的那样。因而,在操作212的网电路中仅仅找回和模拟与用户查询相关的顶端的被预评级的网页,例如根据预评级的顶端1000000个网页。这允许索引中的每个网页的实时查询敏感评级,而不管该索引值有多大。[76] At operation 212, a network circuit is established. As with the circuitry described above with reference to FIG. 1, the web circuitry simulates the index, or a subset of the index. If the ratings are query sensitive, as described above, the circuitry is modified according to the relevance of each web page retrieved by the search to the query. In one embodiment, web pages are pre-ranked according to query-sensitive ratings by first performing operations 212-218, as will be described in detail below. Thus, only the top pre-rated web pages relevant to the user query are retrieved and simulated in the web circuit of operation 212, eg, the top 1,000,000 web pages according to pre-ratings. This allows for real-time query-sensitive ratings of every web page in the index, regardless of how large that index is.
【77】在操作214,如上所述可以调整从索引目录服务出站的类二极管装置以提供依据主题敏感的网搜索,该索引目录服务系与由用户或用户配置文件指示的某个主题相关的。[77] At operation 214, a diode-like device outbound from an indexed directory service related to a topic indicated by a user or user profile may be adjusted to provide subject-sensitive web searches as described above. .
【78】接下来,在操作216中仿真该网电路以确定诸如每个节点的电势(例如电压)。可采用公知的电路仿真软件执行该电路仿真。可选择地,可执行高性能并联多速率电路仿真方法,如2006年12月18日提交的相关美国第11/612,335号专利申请中所述。[78] Next, in operation 216, the mesh circuit is simulated to determine, for example, the potential (eg, voltage) of each node. The circuit simulation can be performed using known circuit simulation software. Alternatively, a high performance parallel multirate circuit simulation method, as described in related US patent application Ser. No. 11/612,335, filed December 18, 2006, may be implemented.
【79】仿真该网电路后,过程执行操作218,其中根据网页相应的节点处的电势来分类网页。如上参考图1所述,每个网页在网电路中被分配一个相应的节点。该仿真将识别每个节点处的电势。具有较高电势的节点将对应于较高质量的网页或更可能包括用户所寻找的信息的网页。随后较高质量的网页被设置为较高级别。[79] After simulating the web circuit, the process proceeds to operation 218, wherein the web pages are sorted according to the potentials at their corresponding nodes. As described above with reference to FIG. 1, each web page is assigned a corresponding node in the web circuit. The simulation will identify the potential at each node. Nodes with higher potentials will correspond to higher quality web pages or web pages that are more likely to contain the information the user is looking for. Web pages of higher quality are then set to higher ranks.
【80】如果在操作210中没有获得用户查询,随后可以执行操作220来获取用户查询。例如,如果该网搜索以非查询敏感模式而被实施,则在模拟因特网之后在操作220中可获得该查询,以获得普通评级。普通评级是不受查询内容影响的评级。一旦获得该查询,仅那些与查询相关的网页从操作218中获得的网页的分类列表中被取出。[80] If the user query was not obtained in operation 210, then operation 220 may be performed to obtain the user query. For example, if the web search is performed in a non-query-sensitive mode, the query may be obtained in operation 220 after simulating the Internet to obtain a general rating. Normal ratings are ratings that are not affected by the content of the query. Once the query is obtained, only those web pages that are relevant to the query are taken from the sorted list of web pages obtained in operation 218 .
【81】在操作222,生成与用户的查询相关的网页的分类列表,并呈现给用户。如果用户正在通过因特网浏览器访问搜索工具,则将输出格式化为HTML的格式并传输到用户的浏览器。[81] At operation 222, a categorized list of web pages relevant to the user's query is generated and presented to the user. If the user is accessing the search facility through an Internet browser, the output is formatted in HTML and transmitted to the user's browser.
【82】如果采用自适应的搜索,则用户在操作224中能够通过除去不期望的结果及返回操作212以建立并随后仿真如上所述的减小范围的修改型网电路来修改搜索结果。否则,如完成框226所示,该过程结束。[82] If an adaptive search is employed, the user can modify the search results in operation 224 by removing undesired results and returning to operation 212 to build and then emulate a reduced-scope modified network circuit as described above. Otherwise, as indicated by completion block 226, the process ends.
【83】广告系统的实现[83] Realization of advertising system
【84】在网搜索期间递送广告的情况下,为了递送相关广告,找到其意义与查询字词相关的字词是重要的。通过将如上所述的仿真方法应用到诸如维基百科(Wikipedia)等在线百科全书,可以找到这些相关的字词。[84] In the case of delivering advertisements during web searches, in order to deliver relevant advertisements, it is important to find words whose meaning is related to the query terms. These related words can be found by applying the simulation method described above to an online encyclopedia such as Wikipedia.
【85】图5示出表示用于实施网页评级系统以识别与查询相关的字词和短语的示意性过程的流程图250。如开始框252所示,该过程开始,并且执行操作254,其中获得一个查询。该查询可以是字词或短语。[85] FIG. 5 shows a flowchart 250 representative of an exemplary process for implementing a webpage rating system to identify words and phrases relevant to a query. The process begins, as indicated by start block 252, and proceeds to operation 254, where a query is obtained. The query can be a word or a phrase.
【86】在操作258,识别与该查询相关的在线百科全书的页面。在一个实施例中,解析该查询以识别所确认的字词或短语。所确认的字词语或短语可包括在线百科全书中具有专用页面的字词和短语。如果找到这样的专用页面,则其成为所确认的网页。如果没有被查询的字词或短语得到确认,则可以识别最接近的匹配,例如包含可能被识别的查询的字词或短语的百科全书的页面。[86] At operation 258, pages of the online encyclopedia relevant to the query are identified. In one embodiment, the query is parsed to identify identified words or phrases. The identified words or phrases may include words and phrases that have dedicated pages in online encyclopedias. If such a dedicated page is found, it becomes the confirmed web page. If no queried word or phrase is confirmed, the closest match may be identified, such as a page of an encyclopedia containing the queried word or phrase that may be identified.
【87】在操作260中,生成在线百科全书的网电路模型。该网电路模型将包括代表百科全书每个页面的节点,如上所述将该节点与类二极管装置相连接,每个类二极管装置代表连接百科全书的其中一个页面到百科全书的另一个页面的超链接。此外,通过能量漏极连接将每个节点与地相连。由于该电路模型的所有节点来自一个已知源极,即在线百科全书,在一个实施例中,所有的能量漏极连接具有一个共同的电阻,其能够被调谐以获得该电路最佳结果。代表所识别的网页的节点被连接到能量源极,例如固定电压源。[87] In an operation 260, a network circuit model of the online encyclopedia is generated. The net circuit model will include a node representing each page of the encyclopedia, connected as described above with diode-like devices, each diode-like device representing a hyperlink connecting one of the pages of the encyclopedia to another page of the encyclopedia. Link. In addition, connect each node to ground through an energy drain connection. Since all nodes of the circuit model come from a known source, the online encyclopedia, in one embodiment all energy drain connections have a common resistance that can be tuned for optimal results for the circuit. Nodes representing identified web pages are connected to an energy source, such as a fixed voltage source.
【88】在操作260中,采用计算机系统来仿真该网电路模型,以计算每个节点处的所期望的电压。该仿真如上所述可被充分地实现,或者采用现有电路仿真软件或者采用如2006年12月18日提交的相关的美国第11/612,335号专利申请中所述的高性能并联多速率电路仿真方法。[88] In operation 260, the network circuit model is simulated using a computer system to calculate the expected voltage at each node. This simulation can be adequately implemented as described above, either with existing circuit simulation software or with high performance parallel multirate circuit simulation as described in related U.S. Patent Application Serial No. 11/612,335, filed December 18, 2006 method.
【89】在操作262中,根据相应节点的电压来对网页进行评级,最高级别的网页对应于最高电压的节点。通过仿真代表百科全书的网电路,可找到与特定查询字词具有相对高的相关度的字词或短语。[89] In operation 262, the web pages are ranked according to the voltage of the corresponding node, the highest ranked web page corresponding to the node with the highest voltage. By simulating a web circuit representing an encyclopedia, words or phrases that have a relatively high degree of relevance to a particular query term can be found.
【90】在操作264中,从多个广告中选择与具有高相关度的字词或短语有关的广告。广告的相关度可以基于广告本身包含的字词或与广告相关的关键字,例如由支付广告费用的人选择的字词,其可以被称为“关键字”。[90] In operation 264, an advertisement related to a word or phrase with a high degree of relevance is selected from the plurality of advertisements. The relevance of an advertisement may be based on words contained in the advertisement itself or keywords related to the advertisement, such as words chosen by the person paying for the advertisement, which may be referred to as "keywords".
【91】在操作266中,生成包含广告的文档。网电路模型中的节点的高电压将指示具有高相关性的词语或短语。包含或与这些相关字词或短语(如通过关键词)相关的广告可被传递给用户。在生成并传递该文档后,如完成框268所示该过程结束。[91] In an operation 266, a document containing the advertisement is generated. A high voltage for a node in the network circuit model will indicate a word or phrase with high relevance. Advertisements containing or relating to these related words or phrases (eg, via keywords) may be delivered to users. After the document is generated and delivered, the process ends as indicated by completion block 268 .
【92】可以理解,虽然示出了因特网网页和电子集成电路之间的模拟,但也可以应用其它具有同等效果的模拟。因此术语“网电路”(“webcircuit”)并不限于电子电路,还包括任何类似的系统。例如,可以使用泵浦和循环水的流水线系统(pipingsystem)进行等同的类比,用阀门代替可变的电阻,用类似单向阀装置代替类二极管装置,且用液压代替电压电势,即势能的另一个形式。因此,虽然该类比有助于理解此处描述方程的非线性系统的概念和数学,用于为评级网页目的而对因特网建模,但是可以跳过生成表示网电路的电子电路的网络表的步骤,而且可以如上所述从网页、它们的内容分数和(诸如加权入站和出站链接的)其它信息直接地生成这些方程式。因此,“网电路”可从概念上被理解为相关节点的系统,每个节点代表一个网页。[92] It will be appreciated that while an analogy between an Internet web page and an electronic integrated circuit is shown, other analogies may be applied to equivalent effect. The term "web circuit" is therefore not limited to electronic circuits, but also includes any similar systems. For example, an equivalent analogy can be made using a pump and circulating water pipeline system (pipingsystem), replacing a variable resistance with a valve, replacing a diode-like device with a device similar to a one-way valve, and replacing a voltage potential with a hydraulic pressure, that is, another form of potential energy. a form. Thus, while the analogy is helpful in understanding the concepts and mathematics of the nonlinear system of equations described here for modeling the Internet for the purpose of rating web pages, the step of generating a netlist representing an electronic circuit of a net circuit can be skipped , and these equations can be generated directly from web pages, their content scores, and other information (such as weighting inbound and outbound links) as described above. Therefore, a "web circuit" can be conceptually understood as a system of related nodes, each node representing a web page.
【93】已知上述实施例,应该理解本发明可采用包括存储在计算机系统内的数据等各种计算机实现的运算。这些运算是那些需要物理量的物理操作。通常,虽然不是必须的,但这些量采用能够被存储、传送、组合、比较和其它操作的电子的或磁的信号的形式。而且,这些被执行的操作术语上通常是指例如生成、识别、确定或比较。[93] With the foregoing embodiments in mind, it should be understood that the present invention may employ various computer-implemented operations, including data stored in computer systems. The operations are those requiring physical manipulations of physical quantities. Usually, though not necessarily, these quantities take the form of electronic or magnetic signals capable of being stored, transferred, combined, compared, and otherwise manipulated. Also, these performed operations typically refer to terms such as generating, identifying, determining, or comparing.
【94】此处所述构成本发明一部分的任何操作是有用的机械操作。本发明也涉及用于执行这些操作的装置或设备。该设备能被专门构造以用于所需的目的,或者该装置可以是由存储在计算机中的计算机程序有选择性地激活或配置的用于通用目的的计算机。特别地,可以采用各种用于通用目的的机器,而根据此处的教导写入计算机程序,或者构建更专门的设备以执行所需的操作可为更方便。[94] Any of the operations described herein that form part of the invention are useful mechanical operations. The present invention also relates to apparatuses or devices for performing these operations. This apparatus can be specially constructed for the required purposes, or the apparatus can be a general purpose computer selectively activated or configured by a computer program stored in the computer. In particular, various general purpose machines may be employed, but it may be more convenient to write computer programs in accordance with the teachings herein, or to construct more specialized apparatus to perform the required operations.
【95】本发明也可以具体化为计算机可读介质上的计算机可读代码来实现。该计算机可读介质是任何能够存储数据的数据存储装置,之后其可被计算机系统读取。计算机可读介质的实例包括:硬盘驱动器、网络附属存储(NAS)、只读存储器、随机存取存储器、CD-ROM、CD-R、CD-RW、磁带和其它光学的和非光学的数据存储装置。该计算机可读介质也可被分布到整个网络耦合计算机系统上以存储该计算机可读代码并以分布的方式执行该代码。另外,可在包含计算机代码的电磁载波中编码本发明。[95] The present invention can also be embodied as computer readable codes on a computer readable medium. The computer readable medium is any data storage device that can store data, which can thereafter be read by a computer system. Examples of computer readable media include: hard drives, network attached storage (NAS), read only memory, random access memory, CD-ROM, CD-R, CD-RW, magnetic tape, and other optical and non-optical data storage device. The computer readable medium can also be distributed over network coupled computer systems to store the computer readable code and execute the code in a distributed fashion. Alternatively, the invention may be encoded in an electromagnetic carrier wave comprising computer code.
【96】本发明的实施例可在单个计算机、或采用多个计算机或相互连接的计算机组件上执行。此处所采用的计算机可包括具有其自己的处理器、它自己的内存和它自己的存储器的单独的计算机系统,或提供计算机资源给网络终端的分布式计算系统。在一些分布式计算系统中,计算机系统的用户实际上可访问在大量用户之间共享的组件部分。因此,用户可访问网络上的虚拟计算机,其对于该用户将作为对单个用户常规且专用的单个计算机而出现。[96] Embodiments of the present invention may be implemented on a single computer, or using multiple computers or interconnected computer components. A computer as used herein may include a stand-alone computer system with its own processor, its own memory, and its own storage, or a distributed computing system providing computer resources to network terminals. In some distributed computing systems, users of the computer system may actually have access to component parts that are shared among a large number of users. Thus, a user can access a virtual computer on the network that will appear to that user as a single computer that is regular and dedicated to a single user.
【97】虽然为清楚理解本发明之目的起见,前面已经详细描述,但显然在附加的权利要求的范围内可实施某些变化和修改。因此,本发明的实施例可被认为是说明性的而非限制性的,本发明并不限于此处给出的细节,而可在附加的权利要求的范围内及等价的范围内做出修改。[97] Although the foregoing description has been described in detail for purposes of clarity of understanding of the invention, it will be apparent that certain changes and modifications may be practiced within the scope of the appended claims. Accordingly, the embodiments of the present invention are to be considered as illustrative and not restrictive, and the invention is not limited to the details given here, but may be made within the scope and range of equivalents of the appended claims Revise.
Claims (28)
Applications Claiming Priority (5)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US79312306P | 2006-04-18 | 2006-04-18 | |
| US60/793,123 | 2006-04-18 | ||
| US11/612,335 | 2006-12-18 | ||
| US11/612,335 US7783465B2 (en) | 2005-12-19 | 2006-12-18 | Parallel multi-rate circuit simulation |
| PCT/US2007/009410 WO2007123919A2 (en) | 2006-04-18 | 2007-04-17 | Method for ranking webpages via circuit simulation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101473304A CN101473304A (en) | 2009-07-01 |
| CN101473304B true CN101473304B (en) | 2016-04-27 |
Family
ID=40829554
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200780022679.2A Active CN101473304B (en) | 2006-04-18 | 2007-04-17 | By the method that circuit simulation is graded to webpage |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101473304B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102663054B (en) * | 2012-03-29 | 2015-08-12 | 天津奇思科技有限公司 | A kind of method and device determining weight of website |
| CN114925308B (en) * | 2022-04-29 | 2023-10-03 | 北京百度网讯科技有限公司 | Web page processing methods, devices, electronic equipment and storage media for websites |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6530065B1 (en) * | 2000-03-14 | 2003-03-04 | Transim Technology Corporation | Client-server simulator, such as an electrical circuit simulator provided by a web server over the internet |
| US8069075B2 (en) * | 2003-03-05 | 2011-11-29 | Hewlett-Packard Development Company, L.P. | Method and system for evaluating performance of a website using a customer segment agent to interact with the website according to a behavior model |
| US7441219B2 (en) * | 2003-06-24 | 2008-10-21 | National Semiconductor Corporation | Method for creating, modifying, and simulating electrical circuits over the internet |
-
2007
- 2007-04-17 CN CN200780022679.2A patent/CN101473304B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN101473304A (en) | 2009-07-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7483820B2 (en) | Method for ranking webpages via circuit simulation | |
| Tanudjaja et al. | Persona: A contextualized and personalized web search | |
| Serdyukov et al. | Modeling multi-step relevance propagation for expert finding | |
| CN101641697B (en) | Related search queries for a webpage and their applications | |
| US6636848B1 (en) | Information search using knowledge agents | |
| Xue et al. | Optimizing web search using web click-through data | |
| Bilenko et al. | Mining the search trails of surfing crowds: identifying relevant websites from user activity | |
| Guan et al. | Personalized tag recommendation using graph-based ranking on multi-type interrelated objects | |
| Xu et al. | Exploring folksonomy for personalized search | |
| US7428533B2 (en) | Automatic generation of taxonomies for categorizing queries and search query processing using taxonomies | |
| CN101460949B (en) | Index documents for information retrieval | |
| EP1596313A2 (en) | Method and system for schema matching of web databases | |
| Baeza-Yates | Web usage mining in search engines | |
| WO2006007229A1 (en) | Method and apparatus for retrieving and indexing hidden web pages | |
| CN101268464A (en) | rank function using document usage statistics | |
| Keenoy et al. | Personalisation of web search | |
| CN102722499A (en) | Search engine and implementation method thereof | |
| Shamsfard et al. | Orank: An ontology based system for ranking documents | |
| Yang | Information retrieval on the web. | |
| Chen et al. | A unified framework for web link analysis | |
| US20090319565A1 (en) | Importance ranking for a hierarchical collection of objects | |
| CN101473304B (en) | By the method that circuit simulation is graded to webpage | |
| US20030195944A1 (en) | Systems and methods for assessing user success rates of accessing information in a collection of contents | |
| Upstill | Document ranking using web evidence | |
| Lei et al. | Improved relevance ranking in WebGather |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| ASS | Succession or assignment of patent right |
Owner name: SYNOPSYS INC. Free format text: FORMER OWNER: GEMINI DESIGN TECHNOLOGY INC. Effective date: 20110505 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20110505 Address after: California Applicant after: Synopsys Inc. Address before: American California Applicant before: Gemini Design Technology Inc. |
|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |