AKS: Informative ranking of sentences and documents

AKS analyzes your search and gives a score for each identified document and sentence in the selected set, based on the amount of information contained in the document or sentence. Thus, one can easily analyze larger document sets because the most relevant documents are retrieved first. Reviewing these documents is often enough, because they contain all the information you are looking for.

Example
This examples drives through an analysis to extract the most relevant documents and sentences out of the search results.

Traditional search engines retrieve references to entire documents containing the search argument, doing so in no particular order. Instead and once it has carried out a conceptual search, AKS provides not only with documents, but also with sentences containing the biological entities of the searched argument (to learn more about conceptual search see Powerful search capabilities oriented to biomedical needs).

In this specific example, we are interested in the relationship between Alzheimer's disease and the chemical huperzine. A list of documents (their titles) matching the query is displayed [find documents containing any of the terms for "Alzheimer's disease" and also containing any of the terms for "huperzine"]. The documents are ranked by their relevance, a quantitative measure of the biological meaningfulness computed on a statistical basis. Therefore, the higher the relevance value, the more likely a document contains valuable information about what we are looking for. If we select the second document from the list, we obtain an interesting and clear summary of the relationship Alzheimer-Huperzine (Fig. 2).

Even more detailed information can be obtained from the sentences containing the biological entities we are interested in. As displayed in Fig 3, the sentences are also shown in a list ranked by their relevance. Again, simply by reviewing the top-scored sentences (even by only reading the first one) we get a global view of the Alzheimer-Huperzine relationship.

Fig 1. List of top-ranked documents from the analysis of the documents containing "Alzheimer's disease" and "Huperzine". The column on the right shows a statistical value of the biological significance of each document. We can click on the second reference to read the complete abstract.

Fig 2. Abstract of the selected document. The bioentities and relevant terms are automatically highlighted (the color indicates the type of bioentity: disease, gene, chemical substance) by AKS, in order to facilitate their recognition and determine the document's topic. By reading this abstract, it is possible to obtain a global view of the Alzheimer-Huperzine relationship.

Fig 3.List of the top-ranked sentences returned by the analysis of the documents containing "Alzheimer's disease" and "Huperzine". The column on the right shows the relevance statistical value of each sentence. The first sentence (the most relevant) is a true summary of the relationship between Huperzine and Alzheimer: chemical features, where it was obtained from, biochemical mechanism of action and use in the treatment of Alzheimer. The remaining sentences confirm the relationship and suggest new relationships to analyze.