Menu
Home
People
Places
Arts
History
Plants & Animals
Science
Life & Culture
Technology
Reference.org
Multimedia information retrieval
open-in-new
<p>Multimedia information retrieval (MMIR or MIR) is a research discipline of <a href="/facts/Computer_science/lN3pEyPz">computer science</a> that aims at extracting semantic information from <a href="/facts/Multimedia/PtJteq27">multimedia</a> data sources. Data sources include directly perceivable media such as <a href="/facts/Content_(media_and_publishing)/eW5vd3Dr">audio</a>, <a href="/facts/Image/fY9DudpT">image</a> and <a href="/facts/Video/F1AV8ben">video</a>, indirectly perceivable sources such as <a href="/facts/Written_language/QQ5DlfoK">text</a>, semantic descriptions, <a href="/facts/Biosignals/PWTD9CWj">biosignals</a> as well as not perceivable sources such as bioinformation, stock prices, etc. The methodology of MMIR can be organized in three groups: </p> <ol><li>Methods for the summarization of media content (<a href="/facts/Feature_extraction/BSMX5e8I">feature extraction</a>). The result of feature extraction is a description.</li> <li>Methods for the filtering of media descriptions (for example, elimination of <a href="/facts/Data_redundancy/QIaooMjL">redundancy</a>)</li> <li>Methods for the <a href="/facts/Categorization/R6rWYttP">categorization</a> of media descriptions into classes.</li></ol>