The following is part of a write up of my readings on semantics in artificial intelligence, specifically in the areas of question answering and document summarization. I'll be updating the post quite regularly until it is done. I am posting here to help me produce small amounts of coherent work.
There are potential benefits to be gained from improving the ability of artificial intelligences to handle semantics. Consider an artificial intelligence in the area of natural language processing. One of the major benefits from such an artificial intelligences is increased flexibility. Such an artificial intelligence upon encountering an unfamiliar situation can use the cues in the meaning of words to seek additional information.
If there are potential benefits then what related work has there been? There has been three main phases of semantics research in artificial intelligence. Semantics from the beginning of artificial intelligence was researched vigorously. A number of challenging issues and more immediately useful non-semantic techniques were discovered. This lead to less research in semantics and more research in areas such as syntactics and statistics. This time a number of issues have been discovered that indicate that semantics may be of assistance. This has led to a resurgence in research in semantics in artificial intelligence.
There has been various types of semantic systems in research over the years. An example of phase one semantic systems includes artificial intelligence systems that operate in a micro-world with a small ontology. An example of phase two semantic systems includes artificial intelligence systems that operate in small domains with a reference corpora. An example of phase two semantic systems includes artificial intelligence systems that operate in large domains with large knowledge databases.
Just as there have been various semantic systems there has been different goals for these systems and different definitions of key terms. System goals may range from matching text fragments to a given query taking into account words that are commonly associated with each other to systems that seek to derive lessons from fables. Definitions of the key term semantic has been used to describe systems upon seeing a key word perform a pre-programmed response to systems that upon seeing a key word deliberates upon the deeper meaning of the word in the wider context of the current task before taking action.
It is important to keep in mind the various goals and definitions in the following discussion of research that supports the point that there are potential benefits to be gained from improving the ability of artificial intelligence systems to handle semantics. It is important to keep in mind in order to reflect on related work and to discern an appropriate direction in which to proceed upon researching. The following discussion will be on the key distinctions, the relevant work and appropriate directions for further research.
No comments:
Post a Comment