Match workflow

The idea is to limit the evaluation of the cartesian product to at most a few initial steps in the match workflow, e.g., one matcher, and to eliminate all element pairs with very low similarity from further processing since they are very unlikely to match. This idea is especially suitable for workflows with sequential matchers where the first matcher can evaluate the cartesian product, but all highly dissimilar element pairs are excluded in the evaluation of subsequent matchers and the combination of match results. 

Peukert et al. (2010a) propose the use of filter operators within match work-flows to prune dissimilar element pairs (whose similarity is below some minimal threshold) from intermediate match results. The threshold is either statically predetermined or dynamically derived from the similarity threshold used in the match workflow to finally select match correspondences. Peukert et al. (2010a) also propose a rule-based approach to rewrite match workflows for improved efficiency, particularly to place filter operators within sequences of matchers. 

Optimizing complete match workflows is so far an open challenge, especially since most match systems prescribe workflows of a fixed structure, e.g., regarding which matchers can be executed sequentially or in parallel. As discussed in Sect. 3.1, (Peukert et al. 2010a) propose a first approach for tuning match workflows focusing on reducing the search space for improved efficiency. 

Coma++ is available for free for research purposes, and hundreds of institutes worldwide have used and evaluated the prototype. Surprisingly, the default match workflow of Coma++ (combining four metadata-based matchers) proved to be competitive in many diverse areas, particularly for matching XML schemas (Algergawy et al. 2009), web directories (Avesani et al. 2005), or even meta-models derived from UML (Kappel et al. 2007). Coma- — - successfully participated in the ontology matching contest OAEI 2006. 

These developments feed off each other. In an integrated workflow the constituent components should be throughput matched in a statistical sense, and as different steps become bottlenecks they are where improvements in throughput should happen. 

Figure 2. Workflow of an LC-MS/MS experiment. A mixture of peptides from a protein sample digest is separated by reversed-phase chromatography on a nano-flow HPLC. The peptides elute from the RP column and are ionized by an electrospray source. In the first stage of mass spectrometry, m/z values and charge states for each precursor ion are determined and the most abundant precursor ions are selected for analysis in the second stage. The ions are then fragmented with by collision-induced dissociation (CID) a gas to produce fragment ions which are detected. Using the mass (from MS-1) and sequence information (from MS-2) protein sequence databases are searched to provide peptide identifications and protein matches.

FIGURE 9.8 Workflow of the BIOS concept incorporation similarity in the protein and the small molecule world and matching... 

In the following, we first describe a general workflow-like approach to apply multiple matchers and to combine their results. We then discuss approaches for instance-based ontology matching and usage-based matching. 

We need evaluation to help distinguish what features and approaches are useful for particular use cases. We need theories to help explain these differences. Tools encode a workflow process and this process must align with the user s own internal process. By aligning these processes, we will be able to assist rather than hinder the user. We must incorporate a human in the loop, where the human is an essential component in the matching process. Helping to establish and harness this symbiotic relationship between human processes and the tool s automated process will allow people to work more efficiently and effectively, and afford them the time to concentrate on difficult tasks that are not easily automated. 

All these technologies must be carefully chosen to meet the objectives in terms of performance criteria, matching the articulation made among the procedures, processes and workflows. 

---