YOUR GO-TO INTELLIGENCE RESOURCE FOR THE SCHOLARLY PUBLISHING INDUSTRY

Recently, two experts offered their views on how to leverage knowledge graphs to maximize benefits from artificial intelligence (AI). The expert views are summarized below. Jans Aasman, an expert in cognitive science and the CEO of Franz.com, in his guest post, describes how knowledge graphs are the path to true AI. This is followed by views from Emil Eifrem, the CEO and co-founder of Neo4j, who directs our attention to graph technology and how it is the secret sauce AI is missing.

Practical artificial intelligence (AI) relies on two methods: statistical reasoning or machine learning and symbolic reasoning based on rules and logic. While the former method learns by correlating inputs with outputs for progressive pattern identification, the latter approach uses expert, human-crafted rules to apply to particular real-world domains. According to Jans Aasman, an expert in cognitive science and the CEO of Franz.com, the two approaches supplement each other for increasingly higher intelligence and performance levels.

Interlinked taxonomies and ontologies form the core of knowledge graphs (KGs). The knowledge graphs— domain knowledge repositories containing ideal machine learning training data—furnishes the knowledge base for maximum productivity of total AI. Hence, the interplay of the knowledge graph foundation with both the statistical and symbolic reasoning form of AI is critical. Firstly, they augment each other by providing the knowledge for rules-based systems and optimize machine learning training data. The machine learning feedback mechanism improves the graph’s knowledge and the rules, while the output from the rules-based systems provides the knowledge for running machine learning.

Secondly, this process is applicable to any number of horizontal use cases across industries. Many such examples—risk management use cases— can be found in the law enforcement and national security domain. For instance, one can observe terrorists, integrate information and create hypothetical events or scenarios based on probability (determined by machine learning). Subsequently, rules-based systems for security measures are transformed into probabilistic rules-based systems that unveil the likelihood of events occurring and how best to mitigate them.

Click here to read the article.

What is stopping artificial intelligence (AI) from becoming smarter? Is it the inability of computers to understand context? Alternatively, is it the need to fit data into storage in order to enable computers to run any useful AI calculation? Just throwing more hardware is not the only answer for optimizing AI. The problem, according to Emil Eifrem, the CEO and co-founder of Neo4j, is how to get to the right data when it misses one crucial axis – relationships?

If real business insight is based on how things are connected, then graph technology – whose sine qua non is working with relationships –can speed up AI software in two ways. It can optimize AI by throwing the right compute power at it and by introducing an optimized-for-relationship data structure.

Furthermore, graph technology can add the missing nuance — how things are related—to machine learning. This will ensure that it is possible to train models to predict based on how things are connected instead of forecasting based on a misleading, unconnected view of the world.

Click here to read the article.