**2. Synthesis**

The traditional use of the word art means skill [10]. The art or skill of knowledge synthesis relies on the ability to retrieve information from peer-reviewed studies and form an academic literature review, which provides more than an annotated bibliography of summaries [11]. A synthesis, the assembly of parts into a new whole, organizes and interprets the concepts, connections, controversies and constraints of a body of literature, filling in gaps and generating new insights, perspectives, directions and novel explanations about the research topic. Table 1 lists over two-dozen types of knowledge synthesis methods [12].


**Table 1.** Knowledge synthesis methods.

**Table 1.** *Cont.*


Note: Table based on Kastner et al. [12].

Other terms for knowledge synthesis include research synthesis, evidence synthesis, and scientific synthesis. Syntheses can integrate current knowledge to inform policy and best practices, as in systematic reviews and meta-analyses, or syntheses can create new knowledge by combining evidence from a wide variety of sources [37]. The latter type of synthesis for new knowledge generation at the source of the flow of scientific information, from theory to practice, is the topic of this perspective article. The term emerging synthesis has been used to describe the ongoing evolution of newer types of synthesis method in which a wide diversity of quantitative and qualitative findings, data, and research designs are combined together to contribute new knowledge and theory to a research area [38].

Explanatory theories that explain causes and e ffects are qualitatively di fferent from descriptive theories that categorize, organize, and describe phenomena [39]. Hjørland's domain-analysis is an example of a descriptive theory used in library science to organize knowledge according to the specific contents of information within a knowledge domain [40]. I propose that di fferences between explanatory and descriptive theories are similar to di fferences between explanatory and descriptive knowledge syntheses. For example, in addition to its use in systematic reviews and meta-analyses, descriptive knowledge syntheses categorize and organize information in taxonomies, ontologies, encyclopedias, databases, library systems, and complex networks. Furthermore, data mining methods have been used to predict new information in complex networks, such as the prediction of paired protein interactions in a protein database, and the prediction of interactions within social networks [41]. However, these methods have limitations. Interactive predictions in social networks, based on observed structural patterns, were shown to have a low rate of correctness [42]. In addition, the method for predicting protein interactions, based on classification of interaction types, does not explain how predicted interactions function, which must be determined with follow-up studies [43], On the other hand, explanatory knowledge syntheses logically combine concepts to infer new explanatory theories and hypotheses which may lead more directly to new explanatory knowledge.
