Commonsense knowledge (artificial intelligence)

World knowledge describes the general knowledge, skills and experience on the environment and society an individual available. It refers to the information stored in every living organism on the world in which he lives and without which this organism could not survive. The world knowledge allows to classify new facts and to act accordingly, even if detailed information is missing. To the knowledge of the world include, for example, background knowledge and encyclopedic knowledge.

Also in robotics plays world knowledge in this broad sense, a role, since computer despite their superiority in information processing many tasks therefore can not solve, because they do not themselves possess knowledge of the world.

Description of phenomena

Even the simplest organisms included (implicitly, not as conscious) world knowledge in the broadest sense, which determines their reactions, such as the " knowledge" of the existence of time, from directions to the relation of cause and effect to the digestibility a specific temperature range or chemical environment for their own survival, etc. These elementary forms of knowledge of the world have already been described by Aristotle and Immanuel Kant as a priori.

The living organisms acquire ( learn ) their knowledge of the world on the basis of the principle of " trial and error " ( "trial and error " ) and they have the option to save this world knowledge of and bequeath (eg genes, memes, tradition ). The principle of trial and error also includes the theory that plays a role, especially in higher organisms, especially humans. This is not necessarily bound to the consciousness, but can vonstattengehen in other structures.

The largest and most important for the survival part of the world knowledge is stored in plants and animals in the DNA ( genome) and is biologically inherited ( instinct ). Some highly developed creatures, especially human beings, also have also the possibility of psychological heredity ( culture) of information that is linked to the various forms of communication.

Based on the concept of genes in biological inheritance, the concept of the meme was introduced as a theoretical construct for the elements in the cultural inheritance. It refers to elementary information units ( blocks) that are stored only in the information storage of the central nervous system of individuals, particularly in the brain. The memetic traditional world knowledge is directly accessible only to a small part of consciousness. Such culturally passed on information can already find in animals.

To the knowledge of the world of man also includes a part of his acquired through awareness and knowledge by means of conscious learning activity, but not temporarily stored, updated information, the special expertise (eg vocational or academic education, expertise, etc.), encyclopedic knowledge showpiece and also not in writing or digitally stored in libraries knowledge (knowledge of the world).

Background knowledge

Background knowledge is common knowledge and knowledge about the world and about society. In a particular environment, for example in discussions, one can assume it to be known. When I say: " The dog of the neighbor let me sleep tonight not ", then I can assume that the knowledge is shared, that is barking.

Encyclopedic knowledge

Encyclopedic knowledge is knowledge that goes all the world. It is the knowledge that is gathered in encyclopaedias. Of course it is time-bound, such as the knowledge in Johann Heinrich Zedler's " Great Universal complete encyclopedia of all arts and sciences " from 1754.

World knowledge in a certain age class

A very recent specific research topic of pedagogy and developmental psychology provides knowledge of the world of children in relation to their general development (eg knowledge of the world of the small and the preschool child, the Seven Years' etc.) and dar. Here in context to their respective cultural environment it goes almost exclusively to knowledge in the conventional understanding, ie knowledge that is represented in the consciousness of accessible memory areas of the brain.

Knowledge of the world as a problem for computational linguists

One can easily imagine that search engines could work more effectively if they could access to holdings of encoded world knowledge. The systematic encoding of world knowledge in the computer could also, in principle, also provide new avenues for Machine Translation.

World knowledge in machine translation

Who wants to translate a text which must have a lot of knowledge about the world to the text zoom bear. For example, should someone who wants to translate a text in which it comes to ticket machines, be informed about how the bus and rail transport is organized. When translating sentences now appear again many translation possibilities. A human translator will be able to exclude many of the theoretically possible translations because of his knowledge of the world. He will say to himself: " This translation may not be right, because yes it is known that the bus driver must not trigger a ticket. " The computer system will not have this knowledge. All attempts to prepare world knowledge so that it can be used by computers, have led to regulations that have become very rapidly unmanageable extensive.

Furthermore: One can only understand the fifth sentence of a text often, if you have read and understood the preceding sentences. It is always possible that the first sentences have provided evidence without which the information of the following sets can not be loaded correctly. A system for the automatic translation of texts should therefore be able to filter out the information content of the read- sets and to store them internally. However, in this case: attempts to take the information from read texts in a computer system, lead to unmanageable large and complex models.

Even if it were possible to save world knowledge in computer- usable form in models and if it would be possible to write computer programs that pull out and save the most important information from texts - it would still completely unclear how to program a computer system so that the translation process " the right questions ", searches its own database in a meaningful way and place to translation-relevant facts.