Inference mechanism

Using these operators, fuzzy inference mechanisms are then developed to manipulate rules that include fuzzy values. The largest difference between fuzzy inference and ordinary inference is that fuzzy inference aUows partial match of input and produces an interpolated output. This technology is useml in control also. See Ref. 94. 

With the view that a KBS interpreter is a method for mapping from input data in the form of intermediate symbolic state descriptions to labels of interest, four families of approaches are described here, each offering inference mechanisms and related knowledge representations that can be used to solve interpretation problems namely, model-based approaches, digraphs, fault trees, and tables. These methods have been heavily used... 

Figure 26, shown earlier, is a simple form of input mapping called table lookup. A more complicated inference mechanism is illustrated in Fig. 30. Here we see a simple example from a fluidized catalytic cracking unit in which multiple product quality attributes can be explained by multiple operating parameters (Ramesh et al., 1992).

Inference paradigms. The basic inference mechanisms of forward-chaining and backward-chaining must be integrated into a real-time execution environment. 

A sample inference mechanism using these facts (given the natural backward chaining of Prolog) might be... 

The classical example of a complex straight-chain reaction for which the results of the steady-state approximation agree with experimental measurements is the hydrogen-bromine reaction H2 T Br2 2HBr [5]. The inferred mechanism is... 

Fig. 3-5. Inferred mechanism of charge transfer in the Earth s redox field. The redox gradient induces the movement of ions toward the environment in which they are able to transfer charge. Ions move only within the Eh-regime in which they are stable, at the limit of which they pass on charge to other species through redox reactions (from Bolviken and Logn, 1975).

Writing empirical rate laws for reactions whose mechanisms have been established is straightforward inferring mechanisms from empirical rate laws is not. Direct detection or trapping of reactive intermediates provides the strongest evidence in snpport of a proposed mechanism. 

The second aspect is the appearance, derived by the face, the clothes, and the behavior of the person. Everything that makes a person friendly to you is included in the decision process. You will automatically rank the facts for instance, the eyes might be more important to you than the hair. If all patterns are evaluated and ranked, you will summarize the outcomes. The next step is to include a certain probability (e.g., maybe the person did not smile at you but at another person) and to account for certain fuzziness (e.g., you still cannot be sure whether there was an intent behind the smile). Finally, you will conclude with a decision as to whether to turn around or not. The inference mechanism is again something that has been trained and is affected by experience Maybe you will not turn around because you had a bad experience with a similar situation before. External facts are also accounted for (e.g., you do not have the time to talk to the person). All of these influences have to be covered if we want to create a computer system that mimics human reasoning and decision making. 

Most expert systems are developed via specialized software tools called shells. These shells come equipped with an inference mechanism — backward chaining, forward chaining, or both — and require knowledge to be entered according to a specified format. We will have a closer look at several tools for developing expert system in this chapter. 

Thus, the observed trend in lifetimes — (C2H )2 C2H HCl <excited mode, the amount of linear and angular momentum which must be incorporated into fragment motion, and the overall number of product channels available. This inferred mechanism is subject to further tests. These include measurements of product velocity and rotational distributions. Real time measurements of either the population in the initially pumped level or the appearance of product fragments would be of great significance. 

Linear plots of In /c versus AG° are found in some electron-transfer reactions i hat proceed by atom transfer, but the slopes are not 0.5 as predicted for outer-sphere reactions by the Marcus theory. Care thus needs to be taken in inferring mechanisms for redox reactions from LFERs. Other LFERs, such as the I lammett and Bronsted relationships, also have been applied to redox reactions of organically complexed metal ions and similarly lead to linear plots of log/c versus E°xb (see next section). 

Inferencing is controlled by the level of belief or probability attached to objects (,demons>, defined below), which can be examined at any point during system development or consultation. This inference mechanism allows the developer to retain straightforward and precise control over inferencing. 

Much of the information required to reason about the PLFs is carried from the Data Entry and Major Unit Process Evaluation modules. Using an essentially forward-chaining inference mechanism, the system uses information already available from these modules to complete a preliminary assessment of each PLF. The initial assessment collects and analyzes information to conclude whether a factor adversely... 

Facts, rules and quantitative models are the chosen forms of representation for the different areas of knowledge in the field of wool dyeing. But these parts are not independent, they must be tied together, i.e. an inference mechanism is needed. The theoretical concepts of forward and backward chaining in their pure forms turned out to be insufficient to catch fully the expert s knowledge about the connections and interdependencies between the various aspects. Only a careful detailed analysis of the expert s behavior allowed implementation of an adequate inference mechanism, which turned out to be a mixture of several theoretical approaches. 

In LISP, all inference mechanisms are implemented by the user. This guarantees that no constraints exist with respect to the type of inference. A disadvantage, however, is the relatively high workload to program the inference schemes. 

---