Ulrich Schade

From Battle Management Language (BML) to Automatic Information Fusion

Current operations, military as well as disaster relief operations, are executed by coalition forces which have to exchange information effectively. This exchange must be supported by a smart architecture of the underlying network and its systems and by a specific language format used to formulate military communications, i.e., orders, requests and especially reports. Such a language format has to support automatic processing of the communication as well as automatic information fusion. Otherwise, the forces’ headquarters would only accumulate huge piles of data without any chance to analyze them in time, i.e., quickly enough to exploit those fleeting opportunities permitted by the enemy or by circumstance.

Interactive multi-robot command and control with quasi-natural command language

Commanding and controlling a multi-robot system (MRS) is a challenging task with complexity increasing progressively as the number and the heterogeneity of robots in the system increases. A natural language-based interactive command and control system that allows a human operator the ability to focus on mission-critical tasks while directing the activities of the MRS is an essential tool to deal with the complexity inherent in a MRS operation. A controlled natural language is an efficient and intuitive way to interact with systems. We have developed and evaluated such a language for giving robots high level orders as well as for adjusting robot behavior with respect to those high level tasks. As additional support, a scheduling system based on genetic algorithms helps the user by suggesting an efficient schedule for all planned tasks and by keeping track of the resources required. In addition, we have also designed a mechanism to aggregate and fuse the information gathered by the robots so that the aggregated information can be sent to the commander formulated using our controlled language, thus enhancing situation awareness without overwhelming the commander with all the myriad details the multi-robot system gathers.

ALARM for early warning: A Lightweight Analysis for Recognition of Menace

Information overflow is a significant and as yet unresolved problem for military, homeland security and law enforcement. Furthermore, quantity is only one part of the intelligence problem: identifying which pieces of information belong together, and assessing the credibility of not only individual pieces of information but also of their correlations to one another complicates things. Automatically sifting, sorting and fusing information garnered from multiple sources into recognizable patterns of behavior and potential threats would provide a distinct operational advantage. This advantage would be clearly increased if the time needed for processing was close to real-time. A number of systems for deeper analysis of potential threats using technologies such as Bayesian networks exist, but tend to be time-intensive. This paper describes a near realtime solution for first-pass processing of inflowing information to provide early warning of developing threats.

Battle Management Language (BML) As an Enabler

According to its definition, a Battle Management Language (BML) is an unambiguous language used to command and control forces and equipment conducting military operations and to provide for situational awareness and a shared, common operational picture. In this paper, we will argue that the use of a BML enables military units to utilize state-of-the-art IT techniques for todays operations. In particular, we will discuss the benefits of using BML for military communications in multinational endeavors, for the use of simulation systems in staff training, for decision support, and for automatic information fusion.

Uncertainty Estimation in the Fusion of Text-Based Information for Situation Awareness

Rapid correlation and analysis of new and existing information is essential to recognizing developing threats and can easily be considered one of the most important challenges for military and security organizations. Automatic preprocessing (fusion) of information can help to “connect the dots” but arguably as important to the intelligence process is the assessment of the combined reliability of the information derived from diverse sources: how can one estimate the overall reliability of a collection of information which may be only partially correct, and is incomplete, vague or imprecise, or even intentionally misleading? In this paper we present a simple heuristics-based model for fusing text-based information from multiple diverse sources which exploits standardized representation of underlying source information. The strength of the model lies in the analysis and evaluation of the uncertainty at three levels: data (source and content), fusion (correlation and evidence), and model (accuracy of representation).

Standardization to Deal with Multilingual Information in Social Media During Large-Scale Crisis Situations Using Crisis Management Language

One of the most effective uses of social media in crisis situation is for the distribution by emergency services, government agencies, and others of reliable information about the nature of the crisis. The panic and confusion that often occurs in the early stages of a crisis can be minimized through such messaging. The value of this role of social media in such cases cannot be underestimated.

Ontologien als Schlüsseltechnologie für die automatische Erzeugung natürlichsprachlicher Texte

In den neuen Medien vor allem im Bereich des Semantic Web gibt es einen permanent steigenden Bedarf an einzigartigen Inhalten und redaktionellen Texten. Dies betrifft sowohl Medienunternehmen als auch Nicht-Medienunternehmen. Der Grosteil der Texte wird heute noch konventionell von Redakteuren erstellt. Dies gilt auch fur Bereiche, in denen Texte recht schematisch und formalisiert auf Basis von Datenbankinformationen oder Excel-Tabellen erstellt werden. Dieser Beitrag beschreibt ein Ontologie-gestutztes Verfahren zur Generierung naturlichsprachlicher Texte, welches auf die fur das Semantic Web vereinbarten Konventionen zuruckgreift [http://www.w3.org/TR/2004/REC-rdf-primer-20040210/, besucht am 05.10.2014, http://www.w3.org/TR/2004/REC-rdf-schema-20040210/, besucht am 05.10.2014, http://www.w3.org/standards/semanticweb/, besucht am 05.10.2014]. Unter dem Begriff Ontologien sind Wissenswelten oder Unternehmens-Enzyklopadien zu verstehen, in denen durch Hinterlegung von Regeln und Anwendung von Inferenzmechanismen neues Wissen aus Vorhandenem abgeleitet werden kann.

Generating Natural Language Texts

This chapter is about natural language texts generated automatically. We will discuss the motivation as to why these texts are needed and discuss which domains will profit from such texts. During recent years NLG technology has matured. Millions of news articles are generated daily. However, there is potential for higher quality and more elaborated styles. For this purpose, new techniques have to be developed. These techniques need to be semantically driven. As examples, we will discuss (a) the use and the retrieval of background information by following paths in knowledge graphs, (b) how to calculate and exploit information structure, and (c) how to hyper-personalise automatically generated news.

Introduction to Semantic Applications

Semantic applications today provide benefits to numerous organisations in business sectors such as health care, finance, industry, and the public sector. These applications use the semantics of a domain in order to improve usability, correctness, and completeness. Developing semantic applications requires methodological skills, e.g., ontology engineering, quality assurance for ontologies, and licence management. Various technologies are available for implementing semantic applications, e.g., data integration, semantic search, machine learning, and complex event processing. This chapter gives an overview of methodologies, technologies, and corporate use of semantic applications.

Comparing single task assignments control with supervisory control through automated plan generation

The workload of a single operator of a multi robot systems increases with the number of robots in use. Supervisory control is a general idea to solve this issue. In this paper we present an experiment in which we compare single robot control and group robot control. Using single robot control the user must task each robot separately. Using group robot control, he can task the whole group and a planning component generates a plan and observes the execution. We compare performance, errors and workload of the user under single robot control and group robot control conditions.