Generic nodes

The iterative search algorithm is being completely rewritten for the search engine implementation to support the use of generic search queries against a file of Markush structure representations. That is, both the queries and the structures on a file may have nodes represented as variable substructures or certain generic nodes (e.g., heterocycle). 

A powerful feature of MDARC is the system of 22 superatoms (Table 3), most of which are generic node types commonly found and referred to in Markush structures in patents. The acyclic superatoms have attributes for low, middle, or high chain length, and straight or branched topology. The cyclic superatoms have attributes for ring system fusion and saturation. For example, a Markush may specify a particular node as... 

Questel has implemented a very powerful display feature, the view focus conunand (vi fo). This command displays the exact features in the indexed structure that caused the hit, and can be very valuable in ascertaining the relevancy of a particular hit. Often, the parent group in an MDARC record has no discernible chemical groups, and is represented with only a grouping of generic nodes. Likewise, when reviewing an abstract or patent specification that was retrieved by an MDARC search it may not be obvious why that document was selected, and the vi fo command can precisely determine... 

Schmuff has published a comparison of MARPAT and Markush DARC software. To generalize, at this stage both commercially available systems have similar capabilities in terms of retrieval and overall functionality. Both systems have the ability to represent in their databases and queries generic nodes typical of Markush structures in patents. Both systems have translation capabilities between generic node.s and specific nodes in database records and queries, and both systems have the ability to highlight elements in the database record that caused the retrieval. The similarities no doubt reflects the Sheffield connection, which both Questel and STN supported, but also the demands the patent searching community, which the developers of both systems consulted with extensively. 

The NOD command also determines generic nodes and groups. The word generic means that atoms and groups are awarded different values (atoms or groups). During the first step the atom is labelled Gk, in a second step the atom is evaluated by the command VARIABLE (VAR) (Fig. 109). 

Table 8 shows the definition of different symbols for generic atoms and shortcut symbols for groups. You can use them in combination with CSS or SSS in order to expand an exact structure. Unlike the generic nodes generic groups cannot be neglected. 

The APS system is supported by an independent modem data model that is often based on generic input/output nodes and their dynamic combination in complex networks and thus better suited for algorithmic processing and optimization than the transaction-oriented business data model of the ERP system. 

By a natural extension of the concepts developed in the previous chapter (structural estimability), if the generic rank of the composite matrix (Aj A2) is not less than n (n number of unmeasured variables), then the system does not include structural singularities. Furthermore, if all the unmeasured nodes are determinable, then there are no isolated variables, which cannot be computed from the balance equations. 

Now, the generic rank of (Ai A2) is equal to the number of unmeasured process variable and the system does not exhibit generic rank deficiency. However, from the corresponding signal graph of Fig. 5, we can see that node 7 is nonaccessible. 

On the other hand, when stream 6 is considered measured instead of stream 3, all of the unmeasured nodes are accessible and the generic rank is equal to 3. 

This example is particularly instructive, especially for the case shown in Fig. 5, since in this case we have no generic rank deficiency but one node is nonaccessible. In such cases we will have overmeasured process variables. This can be seen from Fig. 5, where variable /2 is measured but can also be computed from the balance around unit 1. Note that Equation 1 could also be assigned to solve for variables... 

The data of Table 5.6 were by design collected to have a tree type or so-called hierarchical/nested structure. Figure 5.15 shows a diagram of a generic hierarchical structure for balanced cases like the present one, where there are equal numbers of branches leaving all nodes at a given level (there are equal numbers of determinations for each specimen and equal numbers of specimens for each casting). 

The process models predicting bulk physical attributes have been augmented by a properties model describing the relations between the physical attribues and the sensory attributes. For each of the sensory attribute the lowest complexity model has been determined. Instead of an trial-and-error approach, a neural network with one hidden layer has been used as a generic non-linear function. The complexity of the neural network can thus be seen as the number of nodes in the hidden layer. The performance of the property function model obtained by the above described procedure has verified with the validation set and the property function model provides a good estimation of the sensory attributes. 

Velocity and other variables at cell faces can be obtained by employing suitable interpolation practices. Numerous alternative interpolation methods have been developed. Generically, a value of general variable 0 at the cell face can be expressed in terms of two neighboring nodes and one additional upstream node (the need to include an additional upstream node is discussed later in this chapter). For example, the value of 0 at cell face e can be written ... 

Even after the fixed points have annihilated each other, they continue to influence the flow—as in Section 4.3, they leave a ghost, a bottleneck region that sucks trajectories in and delays them before allowing passage out the other side. For the same reasons as in Section 4.3, the time spent in the bottleneck generically increases as (p-p, ) , where p, is the value at which the saddle-node bifurcation occurs. 

After the context has been checked for all possible rule applications, some rules can be applied, others still have to wait for their context. The next step of the algorithm (find unambiguous rule) tries to find a rule application that is not involved in any conflict. The conflicts have already been determined in the construction phase. As any increment may be referenced by an arbitrary number of links as context, no new conflicts are induced by the context part of the integration rules. The generic PROGRES production in Fig. 3.37 finds rule applications that are not part of a conflict. On the left-hand side a rule-node is searched ( 1) that has only one context node and is not related to any overlap node. It has to be related to exactly one half link ( 2) that does not have another rule node. 

Fig. 5.61. Generic graph transformation to instantiate a new entity node...

All of these constraints are ensured by the generic graph transformation shown in Fig. 5.61 to instantiate a new node for a given subtype of the node class Entity. Relationships between entities are instantiated analogously. 

To facilitate rule deflnition, parameters can be set at deflnition time as well. This can be compared to the concept of genericity as contained in some modern programming languages. We plan to allow different kinds of parameters, ranging from integer values, or node types, to complex graph patterns. 

---