**3. Frameworks for temporal knowledge management and knowledge sharing and their applications**

There are several frameworks proposed in the literature for spatial-temporal knowledge processing in multiple domains, we highlight a few in this section. Within social media, researchers in [25] developed a framework for discovering spatial-temporal patterns from Twitter data. The researchers describe temporal representation using triangulation and clustering to capture twitter events and demonstrated their technique using datasets related to civic unrest in Mexico. The researchers indicate that such a framework can be used to accurately predict the temporal evolution of events depicted in tweets, especially in the age of big data. The researchers indicate the central challenge remains on selecting the optimal window for which to characterize a temporal event. Another challenge is the ability to process spatial-temporal data from multiple sources concurrently. Visualization of temporal patterns is also another issue noted by the researchers.

In recognition that spatial-temporal knowledge can be utilized in the analysis of projects that span territories and time frames, Allais and Gobert in [26], proposed a conceptual framework for spatial-temporal analysis of territorial projects. The researchers partially tested the theoretical framework on several territorial projects. Implementation of such a framework within information systems could provide much temporal knowledge that can be integrated into many territorial-based research and projects.

In modeling climate data, Amato et al. [20] introduced a framework of spatialtemporal prediction using deep learning. The researchers demonstrated their framework through the simulation of two case studies to show coherent temporal fields. One case study was in modeling to predict the temperature in a complex Alpine region in Europe.

Integration of contextual information in the spatial-temporal knowledge process is described in [27] where researchers propose a collaborative framework for planning highway projects by the state highway agencies in the development and maintenance of transportation infrastructure. The framework integrates multiple factors from budgeting, spatial conflict analysis, and their classification which allow the development of response actions. The framework was developed to mitigate incompatible information systems, especially in facilitating dynamic contextual information generated. The researchers claimed that the framework was rated very high by expert users with respect to being easy to use, intuitive, and complete.

Using more than textual data in [28] integrates images in spatiotemporal models geared to person re-identification thus combining visual semantic information and spatial-temporal information into a unified framework.

A meta-learning paradigm is detailed in [29] where researchers create a framework for learning about different cities using spatial-temporal network data and claim that knowledge from one city can be adapted in other cities with respect to factors, such as traffic and water quality. Another framework for analysis of water, energy, and food is

### *A Spatial-Temporal Knowledge Management Framework DOI: http://dx.doi.org/10.5772/intechopen.101797*

described in [30] where an engagement of academics, practitioners, and policymakers are engaged in studying to understand barriers to critical water, energy, and food.

On the visualization of spatial-temporal knowledge, researchers have proposed several techniques. The work in [31] proposed a visual analytics platform that supports large space spatial-temporal data by redefining spatial-temporal attributes in a data-intensive computing environment. This allows distributed storage, data reorganization, query distribution, spatial indices, and fetch segmentations.

The rise of mobile technologies has led to vast amounts of location information generated by individuals. This is quite valuable information in knowledge frameworks, but that data contains a lot of personal information raising data privacy concerns. Frameworks have been suggested for tackling spatial-temporal data privacy such as the works in [32] where researchers use anonymization and probabilistic techniques to mask underlying data map. Recent work introduced temporal hierarchies in the combination of differential privacy to protect temporal data attributes when publishing data about patients [33].

All these frameworks use static data and little attention is paid to the urgency of processing data as is generated or captured. As Shekhar et al. note in [19] extracting interesting and useful patterns from spatiotemporal datasets is more difficult than extracting corresponding patterns from traditional numeric and categorical data due to the complexity of spatiotemporal data types and relationships.

As most of the proposed frameworks focus lies on the percentage of accuracy of model predictions with minimal attention paid to the knowledge creation process, they inherit most of the problems that affect existing knowledge systems with an inability to effectively generate and share knowledge that is actionable in real time. In current frameworks, spatial-temporal data generated lacks privacy protection, is plagued with data quality issues, and models generated are mostly not explainable and are mostly biased. Although knowledge generated from static data is important for lesson learning, it is imperative that on-time knowledge is created, processed, and shared for it to have a clear impact in any business domain while maintaining the privacy of data sources and context. This is particularly critical in cyber networks, critical infrastructure, energy, banking, transportation, and healthcare.

Big data frameworks that recognize the need for real-time processing of data streams have been suggested in the healthcare domain [34]. Though the researcher's methods are demonstrated as effective at processing high-frequency data streams from bedside monitors and sensors, they have yet to be integrated with spatial information that would enhance any resulting knowledge.

In the next section, we describe a conceptual framework for spatial-temporal knowledge generation, processing, and sharing that incorporates components for end-to-end real-time processing of spatial-temporal data streams. This is an overarching spatial-temporal framework that embeds privacy by design, integrates data quality modules, models outputs are encapsulated with explainable details as well as seamless knowledge sharing and application. Design principles adopted in this framework acknowledge the volume, variety, velocity, and richness of spatialtemporal data that is generated every second from multiple domains and as noted in [35], new innovative frameworks for spatial-temporal knowledge management that are actionable are needed to improve lives.
