MOOD AND GENDER DIFFERENCE

The Effect Of Mood And Gender Difference On The Semantic Distance Using Text-Change Detection Methods



The Effect Of Mood And Gender Difference On The Semantic Distance Using Text-Change Detection Methods

There are a large number of methods to analyze landscape change ([Lu et al., 2004], [Mas, 1999] and [Singh, 1989]). Mas (1999) separate between three general methods for change detection using satellite images; image enhancement including numerical combination of image bands using e.g. PCA, one-step classification of multi temporal data, and post-classification methods where two independent land cover classifications are compared. One of the most commonly used techniques is the post-classification approach ([Foody and Boyd, 1999] and Lu et al., 2004 D. Lu, P. Mausel, E. Brondízio and E. Moran, Change detection techniques, International Journal of Remote Sensing 25 (12) (2004), pp. 2365-2407. View Record in Scopus | Cited By in Scopus (217)[Lu et al., 2004]). Given that it is possible to overcome issues of time consumption and required expert knowledge to produce reliable land cover classifications, the main advantages of this method is the detailed information that can be gained from the produced change matrix and the limited impact that image calibration, atmospheric and environmental differences will have on the multi-temporal image comparison (Lu et al., 2004). A further advantage of the post-classification approach is its intuitive interpretation as opposed to numerically based image analysis methods that need careful interpretation to asses what the identified changes mean. This advantage is mainly due to the rich semantics of land cover class labels, but the semantics is also noted by several authors as problematic due to the usually limited descriptions of exactly how land cover labels should be understood (Comber et al., 2004b). In many situations certain types of land use or land cover change is more or less important so that either original classes need to be reclassified into more relevant categories for the analysis or the change analysis on original categories need to be further analyzed and weighted depending on the importance put on certain types of change. Additionally we often find that data on land use and land cover from different times are classified using different classification systems (Comber et al., 2004a). In these situations a standard post-classification change assessment can be very complicated if not impossible.

Many scholars have acknowledged a need to negotiate and compare information from different origins, such as data that use different classification systems. The work on semantic uncertainty (Salge, 1995) and semantic interoperability (Bishr, 1998) of geographic information reflect this concern. Work in computer science, artificial intelligence and information science have also tackled the issue of translations between heterogeneous information sources and many see a potential for using formalized descriptions, or ontologies, that can describe category semantics, to address this ([Fonseca et al., 2002], [Guarino, 1995] and [Sowa, 2000]). The two main approaches to formally represent category semantics that have emerged out of that work use either formal axioms, such as structured taxonomies of classes and sub-classes, or prototypes where objects ...