Qualitative Reasoning and Analysis - Report Example

Qualitative Reasoning and Analysis Qualitative Reasoning and analysis Qualitative reasoning is a branch of Artificial Intelligence. Qualitative reasoning deals with the creation of representations for features that have little information that can aide reasoning. This reasoning reckons that people do not always use differential equations in drawing conclusions concerning the world. As such, people constantly discern the things that happen around them and how to respond to these things without necessarily using quantitative methods. This is the reason as to why human beings are able to build robots that operate in unrestrained environments simulating human cognitions. Engineers and scientists utilize qualitative reasoning to understand a problem and set up ways to help them solve any given problems (Davis, 2008). They also use qualitative reasoning to interpret the outcomes of quantitative measurement, calculation and other functions. Qualitative reasoning helps come up with software that helps scientists and engineers in their work. I have grasped well the concept of qualitative representations. I have understood that there exist many qualitative representations all of which have unique merits and demerits. Qualitative representations avail information to help describe and reason about the aspects of the physical world that are continuous. In qualitative representation, resolution refers to the caliber of information detail in a given representation. Resolution is important in qualitative reasoning as one of the goals of qualitative reasoning is to interrogate the importance of every piece of information in drawing meaningful conclusions. Information that is of low resolution is more readily available than information that is precise (Parsons, 2001). For instance, it is easier to observe that an automobile that is approaching is slowing down or speeding than to derive the automobile’s actual speed. Conclusions that are based on information that is of low resolution are ambiguous. Ambiguity is useful in qualitative reasoning. It helps in predicting future prospects of the observed phenomenon. Individuals use predictions of the future of the phenomenon to decide how much more information they need to collect and depth to which they need to analyze the phenomenon. These future prospects help in deciding the appropriate action based on the options that qualitative reasoning reveals (Shi, 2011). Information of high resolution helps make certain conclusions, but low resolution information helps reveal robust conclusions. Qualitative representation presents knowledge that can easily be taken in by people from different walks of life. Compositionality helps bring together representations for diverse features of an occurrence in order to come up with a whole representation of the occurrence. This is instrumental in formalizing the process of modeling used in qualitative reasoning. A significant number of Artificial Intelligence devices articulate information concerning a given object that carries out few tasks. Engineers and scientists use knowledge that utilizes laws and principles that can be applied variously in regards to the devices they explain and functions in which they can be used (Davis, 2008). Status abstraction is the representation with the lowest resolution. It is used to represent the normality of a quantity. It is classified as the representation that is the weakest of all, and that expresses the functionality of something. The sign algebra expresses whether a continuous parameter is positive or negative. Sign algebra is used in qualitative reasoning concerning dynamics such as the expression of stability and oscillation. It is followed by representation of continuous values using ordinal relations (Parsons, 2001). This is called quantity space representation. For instance, it is used in representing the temperature of a give fluid by considering how the freezing point of its constituent material relates with its boiling point. Quantity space equally supports qualitative reasoning concerning dynamics. While sign algebra will draw values from countable algebraic structures, quantity space will give out resolution that is variable as it allows the addition of new points that can be used to compare. There is a need to improve in the concept of interval representation of numerical values that have a resolution that is variable. Using a set of intervals, quantity spaces can be represented as incomplete information. I need to start at studying the result of comparing the landmark values and the limit points in complete information. Comparing these points defines a number of intervals that serve to partition the value of a parameter. Using this relationship between intervals and quantity spaces in a number of devices that utilize intervals that have known mathematical values can help fine tune qualitative reasoning predictions. Fuzzy intervals help in qualitative reasoning concerning control systems (Shi, 2011). I also have got a lot to catch up on in the order of magnitude representation. This classifies values using a scale of some sort. It is instrumental in unraveling ambiguities and simplification of models. It does this by allowing discernment of what occurrence and outcomes can be ignored in a given phenomenon. For example, in the initial stages of designing a power plant, heat loss through turbines can be ignored as it is negligible in relation to the predicted energy output. Examples of techniques of stratifying values include mathematical thresholds, logarithmic scales and hyper-real values. As I endeavor to seek a better understanding of this concept, I should consider three important things. I have got to consider the point at which a number of seemingly insignificant effects can accumulate to cause significant effects (Davis, 2008). It is crucial to consider the level of soundness of reasoning that I use to support whatever formalism. Thirdly, it is crucial to consider the efficiency of using any given formalism. Qualitative reasoning and analysis have helped me choose and shape my dissertation topic approach. The course has been of particular importance in helping me to understand the concept of behavior. Initially, when I was conceptualizing my dissertation topic approach, I had not understood behavior clearly. The course made this very simple for me to understand as I now know that behavior refers to a set of qualitative states that occur over a given duration. One can use either qualitative or quantitative knowledge or both to describe the behavior. Quantitative continuous parameters bring the idea of trajectory in a state-space model, in the respective numerical features of behavior (Parsons, 2001). The course has also enriched my understanding of histories. Histories define the extent to which a given feature varies in a given duration. The histories of all constituent features and individuals are what make up behavior. In the situation calculus, histories are the dual of individual situations. Histories are bound by space and extend temporarily. Situations, on the other hand, are temporarily bound, but they enjoy global space. Simulation algorithms based on histories are more efficient relative to simulation algorithm based on state (Shi, 2011). This is because simulation algorithms based on histories do not need to consider information that is irrelevant. As regards envisionment, I now know that likely behaviors of physical systems match some arc in the process of envisionment. It is not any arbitrary path in an envisionment can illustrate behavior that is physically feasible. Global constraints are used to test the validity of any such path. Envisionments are used in testing whether observed behaviors are plausible and suggest likely behaviors. However, envisionments are almost always exponential relative to the size of the modeled system (Davis, 2008). Likely behaviors in envisionments are generated like in other fields of Artificial Intelligence. The use of a package has helped me appreciate the research as an easy feasible activity. Before taking this course, I looked at research as one big task which was tedious and boring. I thought the output of research was not worth the strife involved in the process. Data interpretation was one stage in the research process that I thought of as most difficult. Qualitative reasoning and analysis has imparted in me the knowledge of using qualitative state sequences in explaining temporal sequences (Parsons, 2001). It has also given me the skills to interpret the outcomes of successive experiments that use numerical simulation using a phase space qualitative model. This is done by qualitatively describing the physical constraints of a phenomenon. This helps to come up with compatibility constraints that open room for nuance interpretation. When involved in the task of interpreting measurement, one partitions both symbolic and numerical data into intervals. Each of these intervals can be articulated using either qualitative states or sequences of qualitative states. Pre-computed envisionments and on-line limit analysis can help generate likely transitions between states that can be used in data interpretation (Shi, 2011). Compatibility constraints that allow application in either direction are instrumental in providing significant pruning. Examples of constraints include possibilities of certain transitions occurring, approximations of intervals certain states and possibilities of certain states occurring. References Davis, S. (2008). Handbook of Research Methods in Experimental Psychology. Oxford: John Wiley & Sons. Parsons, S. (2001). Qualitative approaches for reasoning under uncertainty. Cambridge, Mass: MIT Press. Shi, Z. (2011). Advanced artificial intelligence. Singapore: World Scientific. Read More