The invention relates to a method for producing correlations and/or interactions and/or knowledge from a plurality of scanned data records. Said method comprises the following steps: a block called a Knowledge Grain (KG) is produced, said block individually providing seed tables comprising concepts and/or terms, each characterising a pre-determined area of knowledge; blocks called Information Grains (IG) are produced, said blocks automatically providing thesaurus tables comprising concepts and/or terms of a scanned data record; vectorial relations (pointers) are created, through which references between identical and/or different grains (IG, IG; KG, KG; IG, KG) can be marked; said pointers are weighted according to consistencies between the grains (IG, KG) related to each other by the pointers; and an action is carried out as soon as a pointer exceeds a pre-determined threshold value.
展开▼