Information structuring for semantic knowledge representation
Below is shown an example of information structuring, which is part of the indexing for semantic search in unstructured text (in this case, online news articles). The news story is about an event (which we here refer to as event #99) where fishermen ‘catch’ $2.5 million worth of cocaine off Florida coast. The extraction of information and the resulting structured information are illustrated by Fig. 1 and 2.
Figure 1: Extract of a news story (left column) and the interpretation of the relevant items (right column) highlighted in the story text. The notation x y tells that x is a kind of y.
- We have no explicit information about the time of the event (when the fishermen found the cocaine), but since June 12 (2013) is the earliest mentioned date, namely that of the Okaloosa County Sheriff’s Office statement, we may assume the event took place the June the 11th or 12th, maybe, but less likely, the 10th or the 9th.
- The location of the event is mentioned somewhat imprecisely in both the Digital Journal article (“about two to three miles off the coast of Destin, Fla., in the Destin East Pass, which is located in the Gulf coast off Florida’s panhandle.”) and in the Okaloosa County Sheriff’s Office statement (“area waters south of the Destin East Pass”). If we can assume that these two pieces of information are independent, we may fuse them to get a more precise location. The fusion would be a kind of confidence weighted intersection of the two imprecise locations.
- The value of the contraband is mentioned in both in the article and in the sheriff’s statement, but since it is the same in all cases (namely 2.5 million USD) the fusion will not change it; besides the article is likely to have applied the sheriff’s statement as source. However, the latter statement tells us more about the kind of value, namely that it is the street value.
Fig. 2 visualizes of the extracted information structure regarding the event (“Event #99”).
This example was included in the organized crime study in José María Blanco and Jéssica Cohen: “Macro-environmental Factors Driving Organised Crime” (Section 4.1 Semantic Search), in Larsen, H.L., Blanco, J.M., Pastor Pastor, R., Yager, R.R. (Eds.). Using Open Data to Detect Organized Crime Threats—Factors Driving Future Crime, Springer International Publishing AG, 2017.