Skip to main content
Figure 3 | BMC Medical Informatics and Decision Making

Figure 3

From: Care episode retrieval: distributional semantic models for information retrieval in the clinical domain

Figure 3

Training the RI-Word model. A sliding window with a size of five words is moved over the text, word by word. The context vector C w 3 ⃗ for the word in the center of the window w3 is updated by adding the index vectors of the other words within the window, i.e. I w 1 ⃗ , I w 2 ⃗ , I w 4 ⃗ and I w 5 ⃗ . As a result, context vector C w 3 ⃗ records the fact that w3 co-occurs with word w1, w2, w4 and w5. The training process continues with moving the sliding window one position to the right and repeating the addition operation for context vector C w 4 ⃗ , and so on until the end of the training text is reached.

Back to article page