Improving the Real-Time Searching in the Organizational Memory
María Laura Sánchez Reynoso, Mario José Diván · Procedia Computer Science · 2019
The real-time data processing constitutes a critical area when talking about real-time decision making. Strong decisions are based on recommendations for describing the associated course of actions, but the real-time processing gives a very short time for searching them. The Processing Architecture based on Measurement Metadata is a data stream engine oriented to measurement projects, which supports the decision making through an organizational memory. The search space related to the organizational memory is initially in-memory limited using the structure of the measurement projects. Given a project, the related projects are ordered based on a given scoring from its structural definition. Here, a new structural coefficient based on the text similarity, which is computed from the textual definition of each descriptive attribute of a project is introduced. This allows better scoring of the related projects, even when its definitions could be affected by human errors or multiple definitions. The scoring is critical when in a given situation, a project has not specific experience for recommending, in such context, the recommendations from the near projects are served. The pabmm_sh library is outlined and a simulation on its associated processing times for the similarity computing are introduced based on the token definition for a measurement project. The library adds a new alternative perspective in the processing architecture for driving the searches into the organizational memory. It can update 2000 projects less than 1 second, keeping the individual processing time of each project under 1 millisecond.