Secure cumulative knowledge memories, immutable transactional log research & development since 1985


  Home page

The Accumulog Knowledge Base

Basic principles

Accumulogs are permanent immutable records of transactions accumulated from the collection of data input as an intermittent or continuous data stream. Unlike the blockchain, which emerged a decade later and which has a rigid convergent data set linked to crypto currency transactions1), Accumulogs qualify data elements by adding an associated elements containing Locational-State2 information. Every dataset element is linked to a location (in geographic space and time) and state (object dataset with object and property dimensions) all recorded at the time of measurement as data related to some target variable and a validation procedure.The object properties section can also include an object method or a link to an algorithm.

Transparency & reliability

Accumulogs were designed to support human and machine learning and decision-making.

Background to this development

Accumulogs are a member of a triad of procedures developed to improve the development and operation of advanced knowledge engineering or AI procedures, created by Hector Wetherell McNeill a British syatems engineering economist and consisting of:
  • Locational-State Theory;
  • Data Reference Models;
  • Accumulogs.
Locational State Theory

As a Senior Scientific Officer at the Information Technology & Telecommuniations Task Force (ITTTF) in 1985, McNeill identified the future potential online problem of "erroneous data" or "false news" delivered remotely becoming a threat to constituent knowledge and democratic decision making and to business economic and financial decision analysis on resources allocation. Therefore the quality or reliability of this information had to be of a high order. Access protocols and information presentation dialogs needed to be designed to facilitate recall on the part of users or machines by providing context as well as exposing inter-relationships within the dataset.

Complete and incomplete data sets

The fact that most datasets do not carry locarional-state information leads to the concept of datasets being complete or incomplete. Complete datasets become more self contained or coherent providing more reliable statistical ANOVA (analysis of variance) results than incomplete datasets.

Context

Context is the logical setting of any analytical environment where the specific interrelationships between objects, be these relationships between physical objects and words or numbers. This relates to the distinction between general semantics and the semantics of words describing the relationships. Without context, AI machine learning and responses become seriously degraded. This was very evident as a result of analysis of automatic translation systems evaluated at the ITTTF, such as SYSTRAN, and which without the system knowing the context of a text, such as "economics" or "philosophy", the translations became inaccurate and sometimes inverted the sense of sentences. Locational-state analysis and Data Reference Models (see below) therefore provided a generic means of preventing models running without context or a method. The role of the validation procedure is to introduce a data reliability and quality control check.

Validation depends upon a dual capacility.
  • The ability to specify data requirements precisely;
  • The ability to assess the degree to which data received remotely oronline meets the established specifications.
DRMs-Data Reference Models

Data Reference Models are a simple procedure to specify data sources, where the sources are located and how the data, subjected to locational-state tags, can be aggregated into algorithms (the object method) to generate useful output. Although applying the DRM appoach since 1985, McNeill only published a short paper on DRMs at the George Boole Foundation Decision Analysis Initiative Workshop in 2014, "McNeill, H.W., "Improving communications within systems groups", Decision Analysis Initiative 2010-2015, Portsmouth, August, 2014". McNeill published this paper to assist government organizations who were attempting to manage "big data" transitions recommended by EUROSTAT and the use of "Data Warehouses" as a basis for avoiding duplication of financial expenditures on additional surveys by taking advantage of existing "administrative data series" colleted for various objectives.

McNeill ascertained that the often general ignorance in government administrations of the actual data sources and methods and timing of data collection by third parties (locational-state references or tags) usually led to a depreciated ability to combine data from such disparate sources in the name of economy. This type of imprecision arose from many different survey designs being dimensioned and stratified according to specific target populations and specific purposes esulting indifferent levels of coherence between surbey errors in representation. The potential costs in terms of inaccuracy and false associations of correletions creating nonexistent cause-and-effect relationships was often too high. This risk was particularly the case in the generation of statistical series to be used in economic policy decision making.

OQSI-Open Quality Statndards Initiative

Locational-State tags, Data Reference Models and Accumulogs have all beenaccepted as recommended procedures by the Open Quality Stadards Initiative (OQSI). DRMs have been widely applied in the agricultural statistical sectors in Eastern Europe and international environmental groups after having been introduced by McNeill.



Hector W. McNeill
Hector McNeill, the originator of the Accumulog and several other procedures supportive of AI is a graduate of Cambridge and Stanford Universities where he studied agriculture, agricultural economics, economics and systems engineering. He initiated this specific development in 1985 as a Senior Scientific Officer at the ITTTF (Information Technology & Telecommunicaions Task Force) of the European Commission in Brussels as the leader of the DELTA programme. This strategic initiative was to advance European capabilities in the development of "continuous learning" for humans, business and in particular knowledge engineering, operations research or what is now referred to as artificial intelligences (AI). This was part of a broader initiative including the multi-billion Euro ESPRIT programme in applied research in response to the 1982 Japanese ICOT Report declaring the intent of Japan to become world leaders in "knowledge engineering" and "5th Generation computing".

McNeill's background included systems engineering economics development including the application of operations research algorithms in agricultural farm planning and policy development. He was the first remote sensing expert of the United Nations Food and Agriculture Organization (FAO) based in Rio de Janeiro. He formed an all-Brazilian systems engineering group to create the world's first fully automatic crop inventory system for the Brazilian Institute of Coffee (IBC). This was financed by the International Coffee Organization (ICO). As a result of work undertaken at the Stanford University School of Engineering, McNeill was aware that the poor resolution of satellite remote sensing (EarthSat) of about 50 metres which was below state-of-the-art technical capabilities and was the result of the intervention of military agencies. This group had designed an earth resources information system satellite and data networks and applications in 1968-69 (Project Demeter). Just 5 years later the FAO Brazilian systems engineering group effort had reduced the operational resolution down to 10 cm a 500-fold resolution improvement over EarthSat. This made it possible to apply advanced computer-based binary pattern recognition procedures developed entirely by this group which was a major advance over the multispectral procedures favoured by NASA.

In 1968 McNeill had created an economic model for the economy which could accept any number of input factors including energy, creating a much improved model in comparison with the current dominant model (Cobb-Douglas) that only used three input categories and did not include energy. This same model was also used to project ecological and agricultural production functions tracing out important relationships to be embedded in the discipline of Agroecological survey and zoning later adopted by the Food & Agriculture Organization (FAO).

McNeill was a member of the group that established the Superintendency for for Natural Resources and the Environment (SUPREN) for the Foundation Institute of Geogrphy and Statistics of the Federal Government of Brazil. He also assisted the Natural Resources Coordination Agency and the Forestry Institute of the State of Sao Paulo in Brazil in the planning the recovery of the Mogi-Guacu River Basin. This included the design of a river/water balance model, the "McNeill-Serra" Model, which was later validated by the Institute of Geology. He completed a computer-based design for a remote sensing-based forest inventory system.

McNeill created SEEL (Systems Engineering Economics Lab) in 1983 to monitor the advance of technologies associated with global networks (later taking the form of the Internet and W3 network) to establish a decision analysis centre based on the purchased output of the Decision Analysis Group at the Stanford Research Institue (SRI) led by Ronald Howard. In 1983-1984 McNeill was the economic modelling contributor to the pioneering study of the consortium made up of General Technology Systems Ltd (UK) and PROGNOS (Switzerland) on the "Potential Contribution of Advances in Information Technology & Communications to the European Economy". The Commission Management at the ITTTF recognised that they needed to adopt the analytical approach to decision analysis and modelling produced by McNeill resulting in his trasfer to the ITTTF.

McNeill was also the Environmental Economist for the G7 Brazilian Rainforest Trust Fund based at the World Bank where he coordinated agro-ecological surveys in the Amazon regions.
KB_OMS the most recent development

Navatec.com, the original developers, "Navatec Voyager" an advanced browser in 2003, are now collaborating with SEEL to create the first macroeconomic administration system, "Knowledge Based Operations Management System (KB_OMS)" as the first AI-supported admin system supporting Real Incomes Objective Price Performance Policy (RIO3P). This is a distinct productivity-based economic development paradigm developed by McNeill in 1975-1976 to respond to the stagflation crisis in 1973. This was caused by OPEC raising the price of petroleum sevenfold between 1973 and 1983. This work draws on McNeill's multi-factor economic model developed in 1968 as well as the subsequent developments of locational-state theory, DRMs and Accumulogs. The recent events linked to the Israeli/Iran military confrontation have created a chain reaction within the commodity complex markets rresulting in the RIO3P approach (RealIncomes Approach) gaining relevance.

Additional information

Unlike block chains, Accumulogs are not just restricted to transactional information which is generally recorded through automated reference coding (convergent) that supports transactions. Accumulogs data sets include additional extended information input by people introducing the risk of wider margins of error as well as the possibility of intentional misrepresentation (divergence). Therefore, if validation detects data quality deficits (essential dataset omissions) these should be corrected and the change/s logged. However, if a user leaves data on the system which validation has flagged to be erroneous or deficient (incomplete) this fact is also recorded.

This helps make Accumulog data streams more reliable and transparent and thereby conforming to the original objective of Accumulogs and Locational State Theory to support learning systems and decision analysis based on reliable information.

Applications

The potential applications are vast. Because of the original expertise of the developer and focus of professional effort6, Accumulogs have been developed to address one of the most complex decision analysis domains of natural resources, agriculture, innovation and economic development. These domains combine a wide range of exacting technical and economic information with conditions of uncertainty. The general approach is to build decision analysis models of any proposed action such as a project and use Locational State considerations to evaluate the potential impacts of those factors characterized by uncertainty so as to quantify the dimensions of risk as a basis for selecting preferable options for actions or project designs.

Practicality

Although Accumulog and Locational State Theory combine to create novel and sometimes complex concepts, the development of Accumulogs has been directed towards eminently practical applications.

Data streams are generated by:
  • ongoing activity implementation (such as a production system or project)
  • simulations of agricultural development, innovation and economic development processes or projects during their design phases
This data is streamed to an Accumulog that records all data to support:
  • the management of all phases of individual project or process design and implementation phases
  • the management of all projects based on a transparent multi-project or process portfolio operation
All data streams into a Portfolio Data Warehouse (PDW)7. A PDW is created by the configuration of all of the project data sets contained within the Accumulog as a portfolio. This provides a transparent support for donors, investors and project managers. They have total oversight through a real time audit (RTA)8 and real time decision support in response to change. The Data Warehouse model enables the sharing of benchmarks and lessons learned on projects under the management of a single organization. The data acquisition and sharing model is completely decentralized enabling those who generate the data to monetise selected elements. For example supplying operational performance benchmarks achieved by different types of project or business process to agricultural or manufacturing extension services or statistical organizations. As a portfolio is extended through the addition of new projects, the datasets provide an increasingly refined knowledge base on the association of performance benchmarks with specific project level contexts. This increases the value of the information.

The Accumulog model applied to project cycle and portfolio management contains data covering the whole project cycle including design, procurement, implementation and post-funding activity. Part of the OPEE structure is that projects are designed on the basis of deterministic decision analysis models9 so as to enable simulation of project option scenarios including alternative input and output circumstances, processes and techniques and such factors as weather impacts. The system also enables evaluations of potential external and internal change impacts on a project as well as impacts on the communities, the environment and ecosystems. The simulation techniques so far deployed in the Accumulog configurations include the most reliable operations research techniques including Monte Carlo Simulation, SIMPLEX optimization, sensitivity analysis, human resource learning curves, dynamic chain sequences (including Markov chains) and population growth impacts of resource consumption and requirements. Such simulation models are subject to assessment against actual data and benchmarks to ensure reliability for the purposes of project design and in support of implementation decision-making.

Improving the value of historic datasets

Amongst the most difficult data to collect is accurate locational state data, especially that relating to phenomena such as weather patterns, environmental conditions and natural processes or even political and economic events. Standard blockchains are a static ledger. Accumulogs, on the other hand, also collect new data which is validated, and especially locational state data. This enables Accumulog records that were input in the past to be related to increasingly refined locational state data as a basis for improving the understanding of the nature and relationships between data elements input in the past. This provides a powerful vector for learning based on instructional simulation using more refined and complete data. This does not alter the original data records but it enables layers of more in-depth interpretative analysis to be added helping to reveal previously "hidden" relationships. As a result, Accumulogs are not static ledgers in the common sense of these terms but they extend blockchains with additional non-intrusive data that can expose valuable data relationships based on the OPEE approach and simulation. This is consistent with the original purpose of Accumulogs to support learning systems and decision analysis.

References:

1. Crypto-currency transactions on blockchains, such as BitCoin, usually register the identities of transacting parties, direction of transfer, a date-time stamp, the currency ID, quantities involved and exchange rate (price quotation).
2. Locational State theory embeds time-space elements into datasets to secure absolute coordinates of events and transitions. Locational State site
3. Plasma Operating System (POS) is an internal control framework for cloud-based server-side scripts that has been developed for the Plasma DataBase by Navatec.com.
4. Object Profile Elements Extension (OPEE) is an extension to the OOP profile that adds validation as an essential support for Accumulog operations. In terms of scripting it represents an added component to the ECMAScript and ISO international JavaScript server side extension. Now superceded by developments by Navatec.com.
5. Object Oriented Procedure (OOP) (OOP) was developed in the 1960s as a means of scripting reality to build computer-based simultation models such as the SIMULA series. This development took place in Norway. Kristen Nygaard started writing computer simulation programs in 1957. Nygaard saw a need for a better way of describing the heterogeneity and the operation of a system. To go further with his ideas on a formal computer language for describing a system, Nygaard realized that he needed someone with more programming skills than he had. Ole-Johan Dahl joined him on his work January 1962. SIMULA 67 and modifications of SIMULA were used in the design of VLSI circuitry (Intel, Caltech and Stanford). Alan Kay's group at Xerox PARC used SIMULA as a platform for their development of Smalltalk (first language versions in the 1970s), extending object-oriented programming importantly by the integration of a graphical user interfaces and interactive program execution. Bjarne Stroustrup, from Denmark, started his development of C++ (in the 1970s) by bringing the key concepts of SIMULA into the C programming language. The idea of this development arose as a result of his doctorate work at Cambridge University. SIMULA also inspired much work in the area of program component reuse and the construction of program libraries. The central operational implementation strategy in support of Accumulogs and LST is the use of a server side extension of the international standards for JavaScript which supports OOP. As a result the operational framework builds simulation models by default.
6. Developer Accumulogs and Locational State Theory were both conceived by H. W. McNeill in 1985, a Senior Scientific Officer at the Information Technology & Telecommunications Task Force (ITTTF) in Brussels who managed the DELTA learning systems initiative. McNeill is an agronomist with post-graduations in economics and systems engineering from Cambridge & Stanford Universities. He is a specialist in the development of coding methods and procedures for decision analysis applied to agricultural economic development project cycle and portfolio management.
7. Portfolio Data Warehouse (PDW) was proposed by Hector McNeill in 2016 as a more effective substitute for General Data Warehouses or Data Warehouses, sometimes referred to as Big Data. McNeill has argued that PDWs represent a convergent and coherent evolving quality and value of knowledge content enabling the specification of coherent deterministic decision analysis models based on more complete datasets. General Data Warehouses (GDW) combine data collected for different purposes including administrative and regulatory information. This type of information is often modified by data suppliers and even authorities so these inaccurate and unreliable data elements can add too much noise to the system. This often results in a lack of required levels of associative coherence thereby reducing the utility of this information. This reduces its value in specifying good quality deterministic decision analysis models essentialfor human and machine decision analysis. See Agricultural innovation.
8. Real Time Audit (RTA) Real Time Audit is a 24/7 real time oversight system that operates within the Navatec Cloud, with global coverage, to access all projects in portfolios that use the Navatec System, a cloud-based Software as a Service (SaaS) for project cycle and portfolio management.
9. Decision Analysis Models (DAMs) are based on R.Howard's deterministic decision analysis models developed at the Stanford Research Institute. These have been advanced by SEEL (Systems Engineering Economics Lab) to combine OOP and OPEE as a Seel-Telesis Systems Development Programme activity.
10. ADA ADA is an Accumulog Data Access controller introduced by Plasma.Systems in July 2020. This provides a flexible data entry with update logs or without and a conversion to imutability when a specific LST coordinate is reached, such as a date. This prevents any alteration of proposals, contract and transational agreement specifications and conditions. This has been accepted as an OQSI standards recommendation as part of their OQSI 3DP (Due Diligence Design Procedures) for projects.