What are the tasks of link mining?

There are several tasks of link mining which are as follows −

  • Link-based object classification − In traditional classification approaches, objects are classified depending on the attributes that define them. Link-based classification predicts the category of an object depends not only on its attributes, but also on its links, and the attributes of linked objects.

    Web page classification is a well-identified instance of link-based classification. It predicts the classification of a web page based on word appearance (words that appear on the page) and anchor text (the hyperlink words, that is, the words it can click on when it can click on a link), both of which serve as attributes. Furthermore, classification depends on links between pages and different attributes of the pages and links.

  • Object type prediction − This predicts the type of an object, depending on its attributes and its links, and on the attributes of objects connected to it. In the bibliographic domain, it can be required to predict the venue type of a publication as either conference, journal, or workshop. In the connection domain, the same task is to predict whether a connection contact is by e-mail, phone call, or mail.

  • Link type prediction − This predicts the type or goals of a link, depending on the properties of the objects contained. Given epidemiological data, for example, it can try to predict whether two people who understand each other are family members, coworkers, or acquaintances.

  • Predicting link existence − Unlike link type prediction, where it can understand a connection exists between two objects and is required to predict its type, instead it can want to predict whether a link exists between two objects. Examples contain predicting whether there will be a link between two Web pages and whether a paper will cite another paper.

  • Link cardinality estimation − There are two forms of link cardinality estimation. First, it can predict the number of links to an object. This is beneficial, for example, in predicting the authoritativeness of a Web page depending on the number of links to it (in-links). Similarly, the multiple out-links can be used to recognize Web pages that act as hubs, where a hub is one or a set of Web pages that point to several authoritative pages of the same case.

  • Object reconciliation − In object reconciliation, the function is to predict whether two objects are literally the same, based on their attributes and links. This function is common in information extraction, duplication removal, object unification, and citation connecting, and is also referred to as record linkage or identity uncertainty.

Updated on: 25-Nov-2021


Kickstart Your Career

Get certified by completing the course

Get Started