DeJong, G., Prediction and Substantiation : A New Approach to Natural Lan- guage Processing, Cognitive Science 3 (1979) 251-273.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....frames can be coded to understand entire stories [30] or for understanding the constituents of a person s name [16] A frame based system that attempted to detect events on a newswire was constructed by DeJong in the late 1970s. He used pre specified software objects called sketchy scripts [26]. Frames and scripts for general news events such as Vehicular Accidents and Disasters were constructed by hand. The goal of his system was to predict which frame needed to be populated. During processing, a frame was populated with words from the text, and a script was traversed to produce a ....

G. DeJong, "Prediction and Substantiation: A New Approach to Natural Language Processing," Cognitive Science, 3: 251-273, 1979.

....Handling news The tasks defined within TDT, Event Tracking among them, appear to be new within the research community. Some efforts have been made to classify news stories into broad topic areas automatically using nearest neighbor matching[15] pattern matching[9] or frame based representations[7]. For the most part, those techniques are intended to match stories against a set of topic labels that are known a priori. Event Tracking requires finding stories that discuss an event that may not match any already known class of events. Because it connects stories together by the driving event, ....

G. DeJong. Prediction and substantiation: A new approach to natural language processing. Cognitive Science, 3:251--273, 1979.

....Some event related work has been reported prior to the TDT workshop, but new event detection has not been a focus. A frame based system that attempted to detect events on a UPI newswire was constructed by DeJong in the late 1970s. He used pre specified software objects called sketchy scripts [8]. Frames associated with 50 general events were constructed by hand. The goal of his system was to predict which frame needed to be populated, and then to produce a short summary of the event. DeJong s system was primarily a natural language parser that detected when a document contained an event. ....

G. DeJong, "Prediction and Substantiation: A New Approach to Natural Language Processing, " Cognitive Science, 3: 251-273, 1979.

....but inappropriate for detection. Event detection (and to some extent tracking) requires finding stories that discuss an event that may not match any already known class of events. One study that is close in spirit to the TDT work was done by DeJong using frame based objects called sketchy scripts [8]. Frames associated with 50 general events were constructed by hand. The goal of his system was to predict which frame needed to be populated, and then to produce a short summary of the event. DeJong s system was primarily a natural language parser that detected when a story contained an event. It ....

G. DeJong. Prediction and substantiation: A new approach to natural language processing. Cognitive Science, 3:251--273, 1979.

....attached to it. I consider these situations in turn. 3. 2 Knowledge structures are not perfect Conventional script, frame or schema based theories assume that understanding means finding an appropriate script, frame or schema 3 in memory and fitting it to the story (e.g. Cullingford, 1978; DeJong, 1979] Schemas in memory are assumed to be correct in the sense that they are completely understood and constitute a correct model of the domain. If an applicable schema is found, it is instantiated and the story is assumed to be understood. However, this model is inadequate since in practice an ....

....of what it means to understand a story, usually defined in terms of coherence relationships at some arbitrary level of detail. Typical story understanding systems are usually designed either to read stories in depth (e.g. BORIS [Dyer, 1982; Lehnert et al. 1983] or to skim stories (e.g. FRUMP [DeJong, 1979]) However, these systems cannot decide the depth to which the story should be processed or which inferences should be drawn during the understanding process because they do not maintain an explicit model of their learning goals. In other words, such systems are not trying to learn about anything ....

G. F. DeJong. Prediction and Substantiation: A New Approach to Natural Language Understanding. Cognitive Science, 3:251--273, 1979.

....convert patient discharge summaries to a form suitable for use as input to a traditional Conference on Data System Languages (CODASYL) database management system. One of the first reported meaning extraction systems, which operated on texts of unrestricted topics, was implemented by Gerald deJong [DeJong 79] DeJong 82] Using a newswire network as its data source, deJong s program, FRUMP, monitored a newswire using simple scripts designed to cover news stories. FRUMP sought to match each new story with a relevant script on the basis of keywords and conceptual sentence analysis. In 1980, DaSilva ....

DeJong, G. F. Prediction and Substantiation: A New Approach to Natural Language Processing. Cognitive Science, 3, pp. 251-273, 1979.

No context found.

DeJong, G., Prediction and Substantiation : A New Approach to Natural Lan- guage Processing, Cognitive Science 3 (1979) 251-273.

No context found.

G.F. DeJong. 1979. Prediction and substantiation: A new approach to natural language processing. Cognitive Science, 3:251--273.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC