The INTERGATOR SMART SEARCH is a novel enterprise search based on artificial intelligence. Unlike a conventional enterprise search, INTERGATOR SMART SEARCH offers a broader view of the structured and unstructured data scattered throughout the company. An explorative search with a so far unique graphical user interface component – the so-called context map – visualizes thematically grouped search hits and thus shows connections between terms and data. In SMART SEARCH, documents and data are not only prepared in a structured manner in an index, but are also analyzed with neural networks via machine learning and the recognized properties are transferred into a model. In this way, similar terms and search hits can be found more easily and linked semantically and associatively.

SMART SEARCH, Insight Engine and Cognitive Search​

Even though search technologies are becoming more and more sophisticated and more and more systems can be linked to a search, users still often face the dilemma of not knowing how to formulate a concrete query in such a way that a satisfactory result is returned. Unclear search terms, ambiguities or a general lack of knowledge about the available database make an efficient search more difficult. Up to now, data and documents could be found either through the terms contained in them or with the support of elaborately maintained thesauri or synonym dictionaries.

There are various concepts for improving searches with artificial intelligence. What is described as Insight Engine or Cognitive Search also uses AI methods, but the implementation of INTERGATOR SMART SEARCH goes one step further.

INTERGATOR SMART SEARCH also offers a native context-sensitive search. The search terms no longer necessarily have to be contained in the texts to be searched. The innovation of INTERGATOR SMART SEARCH lies in the combination of the keyword-based search with a model-based, semantic-associative search.

Keyword based and intelligent search combined

This new approach offers decisive advantages, since the user is shown not only search hits containing the term but also thematically relevant hits without direct reference to the terms in the search query. The special feature of INTERGATOR is that the model-based, context-sensitive search in a query can be combined with keywords, wildcards and metadata filters (so-called facets) – if required also with Boolean operators and classic query syntax.

The prerequisite for this is a model trained by machine learning from the data stock, which contains a mathematical representation for each document. SMART SEARCH compares terms and content on the basis of this model and thus evaluates the relation of the content to each other and to the search terms. Previous search solutions only extended the search query by additional terms.

This concept becomes particularly clear with ambiguous terms such as the word “bank”. A bank can be a seat, a financial institution, a storage unit or part of a combustion engine (cylinder bank). The technology-savvy user may be able to understand these differences, but previous lexical searches yield stoical results without understanding the context. This is where the approach of Natural Language Processing (NLP) comes into play, in which human language is analysed, interpreted in terms of meaning and the context is recognised. With the help of machine learning, models can be trained to extract essential information from data and documents and to view the information stock in its entirety. If, for example, terms such as money, credit or interest are used in a document, the model locates it more in the direction of finance than, for example, in the subject area of furniture. Through this technology, information can be transformed into knowledge.

Query-by-Example and language independence

INTERGATOR SMART SEARCH searches semantic-associative, cross-lingual, natural-language and can find other documents using example documents or images. The latter, also known as Query-by-Example (QBE), extends the search from single terms to complete text paragraphs or entire documents. The search with such query texts is not based on the terms contained in the texts, but is also performed by a mathematical representation of the text content.

The training of a model with the help of neural networks has another advantage. INTERGATOR SMART SEARCH searches in documents and data independent of language and simply transfers the query to another language area. In this way, foreign-language texts also become part of the hit list without the need for prior translation of the contents.

With INTERGATOR SMART SEARCH you receive a comprehensive solution for up-to-date information and knowledge management. SMART SEARCH makes the knowledge documented in the digital contents faster and more completely accessible.