Pathfinder Networks

Home     Download     Examples    FAQ     Citations   Wikipedia


Here is an example of an undirected Pathfinder network derived from average similarity ratings of a group of biology graduate students. The students rated the relatedness of all pairs of the terms shown, and the mean rating for each pair was computed. The network shown is the PFnet(2, infinity). This picture was generated with the Matlab version of the software.

 

Here is a network [PFNet(2,infinity)]of concepts relevant in aviation derived from the average relatedness ratings of a group of experienced aviators.. It was also drawn using the Matlab version of the software.

Here are two nearest neighbor networks of the same concepts the first from the Java software and the second from the Matlab software. In these networks, arrows point from a node to its nearest neighbor (or neighbors if there are ties). Note that the network is not necessarily connected.

This is another network derived from the same data as the first one. this time the network is PFnet(n-1, infinity). It was produced with the Java version of thw software (JPathfinder.jar). This network has all of the links that would be found in any mimimum spanning tree of the complete network corresponding to the proximity data. In this case there are two minimum spanning trees. One would result from removing the link between "living thing" and "mammal." The other would result from removing the link between "living thing" and "deer." Both links are included in the Pathfinder network because they have the same proximity value. Also note that the first network has all of the links shown here as well as some additional links that are included when the q parameter is equal to 2.

This PFnet(n-1, infinity) is derived from the average ratings from undergraduate students taking an introductory psychology course. Note the differences in the links with people who differ in their knowledge of biology. Links to "mammal" are particularly revealing. Coincidentally, there are also two minimum spanning trees in this case. One would result from removing the link between "living thing" and "bats." The other would result from removing the link between "bird" and "bats." These links are tied the the proximity data.

 

Here is a PFNet(n-1,infinity) of the 20 nearest neighbors of Watergate derived using Reflexive Random Indexing on the Touchstone Applied Sciences (TASA) Corpus.

 

Here is an image of a portion of a very large network derived from analyzing the coocurrence of about 4,000 terms in Longman's Dictionary of Contempory English. The network shows all of the words within 3 links of "bank." It illustrates the ambiguity of "bank" by separating the terms related to the different meanings of the word. This is a pciture from an old, defunct version of the software than ran under DOS. Thus it has some historical value.