This is the first article in a series dedicated to one of the most promising areas in the field of artificial intelligence - decision support systems (DSS). They are also called Decision Support System (DSS) or digital advisors. The second article is available here, the third here.
In the article Artificial Intelligence: Assistant or Toy? We reviewed the key areas for the use of AI:
forecasting and decision making;
analysis of complex data without clear relationships, including for forecasting;
process optimization;
pattern recognition, including images and voice recordings;
content generation.
Areas in AI that are currently at the peak of popularity are pattern recognition, including images and voice recordings, as well as content generation. And this is where most AI developers go. At the same time, this area is the most problematic and almost all developers of such AI solutions (as of October 2023) report losses. But we will look at this in a separate article.
Let's return to our topic. For economics and business, the main direction is forecasting and decision making. And it is in this direction that key investments from the state and investors are going.
So what is a DSS? Where did they come from? What awaits us? Let's look into these.
Content
DSS or digital advisor is software that is used to make decisions in complex situations. The key task is to analyze data in complex conditions and prepare recommendations.
All this magic can work based on technologies such as:
neural networks and machine learning;
big data and end-to-end analytics, lakes and data warehouses;
internet of things;
cloud computing;
digital twins;
game theory, systems limitation theory;
systems of rules based on expert knowledge.
DSS have been around since the late 1960s. More precisely, model-oriented DSS (Model-oriented Decision Support Systems - DSS). Before this, there were simply solutions based on static reports. This is approximately what most Russian companies now use 1C for.
And as a result, the end of the 1960s is now considered the beginning of modern DSS and digital advisors. Let's go through the chronology and key milestones.
1971 - a book was published in which for the first time the results of implementing DSS, based on mathematical models, were described.
1974 – the definition of MIS (Management Information System) was given:
“MIS is an integrated human-machine information system that supports the operations, management and decision-making functions of an organization. The systems use computer hardware and software, management and decision-making models, and a database.”
1975 – criteria for designing DSS in management were proposed.
1978 - a textbook on DSS was published, which describes aspects of creating a DSS: analysis, design, implementation, evaluation and development.
1980 – the basics of the DSS classification were given.
1981 – the theoretical foundations for the design of DSS were created and 4 key components of all DSS were identified:
1) Language system (LS) – the DSS can receive all messages;
2) Presentation System (PS) (DSS can issue its own messages);
3) Knowledge System (KS) – all knowledge of the DSS is retained;
4) Problem-Processing System (PPS) – a software “mechanism” that tries to recognize and solve a problem while the DSS is running.
1981 – describes how a DSS can be built in practice. At the same time, the Executive Information System (EIS) was developed - an IT system that provides the manager with information for making management decisions.
1990s - so-called Data Warehouses are developed.
1993 - E. Codd proposed the term OLAP (Online Analytical Processing) for a special type of DSS - operational data analysis, analytical data processing to support important decision making online. The initial data for analysis is presented in the form of a multidimensional cube , from which you can obtain the necessary reports.
2000s - a Web-based DSS was created.
2005 - at the International Conference “Information and Telemedicine Technologies in Health Care” (ITTHC 2005) in Moscow, A. Pastukhov (Russia) presented a new class of DSS - PSTM (Personal Information Systems of Top Managers). The main difference between PSTM and existing DSS is the construction of a system for a specific person. We apply the same logic in our product - a digital advisor for project management , where the personal qualities of the manager will be taken into account.
DSS can be used in any areas and industries:
decision-making and formation of treatment protocols in healthcare;
calculation of prices and coordination of needs;
supply chain optimization;
making decisions on providing loans;
insurance;
configuration of IT systems;
revenue and profitability forecasts;
prediction of natural disasters and recommendations on measures to prevent or eliminate them;
optimization of product lines, loyalty programs and minimizing customer outflow;
procurement and storage planning, etc.
In general, in the digitalization strategy of almost any large company there is a clause on the creation of a corporate DSS, and this is the next step after the implementation and development of business intelligence solutions (BI systems).
And for solutions of this class, when registering in the register of the Ministry of Digital Development of the Russian Federation, a separate class of software is allocated.
If we talk about strategy at the state level, thenthe Decree of the President of the Russian Federation of October 10, 2019 No. 490 “On the development of artificial intelligence in the Russian Federation” approved the National Strategy for the Development of Artificial Intelligence for the period until 2030. And all promising DSS solutions will be based on AI. At the same time, it is useful to look at the distribution of costs among all end-to-end digital technologies.
In general, AI-based DSS are the most promising direction in the field of digitalization and automation. And the solutions themselves will become the standard for all companies and industries. The popularity of this trend can be judged by the number of articles and materials published in the Russian State Library .
However, such a system can only be created after reaching a certain level of maturity, especially in working with data.
There are generally accepted 2 approaches to the classification of DSS.
1. Based on the data/model relationship (Stephen Alter’s method)
FDS (File Drawer Systems) - systems for providing access to the necessary data;
DAS (Data Analysis Systems) - systems for fast data manipulation;
AIS (Analysis Information Systems) - data access systems according to the type of solution required;
AFM(s) (Accounting & Financial models (systems) - systems for calculating financial consequences;
RM(s) (Representation models (systems) - simulation systems;
OM(s) (Optimization models (systems) - systems that solve optimization problems;
SM(s) (Suggestion models (systems) - systems for constructing logical conclusions based on rules.
2. By type of tools used:
Model Driven - based on classical models (linear models, inventory management models, transport, financial, etc.);
Data Driven - based on historical data;
Communication Driven - systems based on group decision-making by experts (systems for facilitating the exchange of opinions and calculating average expert values);
Document Driven - essentially indexed document storage;
Knowledge Driven is a system based on knowledge, both expert (human) and determined using AI and machine learning.
We got to know what DSS are and why they are needed, when they appeared and how they developed, what they are. In the next article we will take a closer look at:
how the DSS is architecturally designed;
what are the shortcomings and limitations of current DSS (expert and AI-based);
who is responsible for decisions and recommendations, and why this is a key issue for the development of the industry.
Useful materials: