“ Automatic summarization is the process of shortening a text document with software, in order to create a summary with the major points of the original document”

WIKIPEDIA.COM

APPLICATIONS

newspaper

time saving

Approachs

Cosine Similarity

Text Rank

K-Mean Clustering

Credits: Paek Jeongyeup

Credits: Paek Jeongyeup

Credits: Paek Jeongyeup

Credits: Paek Jeongyeup

How to choose relevant sentences

based on distance to cluster centroid

with 3 clusters A, B, C

Evaluation

ROUGE is a score of overlapping words. ROUGE-N refers to overlapping n-grams

an n-gram is a contiguous sequence of n items from a given sample of text or speech.

RECALL

• ROUGE-n recall=40% means that 40% of the n-grams in the reference summary are also present in the generated summary.
• ROUGE-n precision=40% means that 40% of the n-grams in the generated summary are also present in the reference summary.
• ROUGE-n F1-score=40% is more difficult to interpret, like any F1-score.

http://anthology.aclweb.org/W/W04/W04-1013.pdf

Lin, Chin-Yew. "Rouge: A package for automatic evaluation of summaries."

Results

Results

Results

Results

K-mean clustering

10 Clusters

input text: "Artificial Intelligence"

most relevants word per cluster (stemmed)

GUI

Improvement axis

• Acronym replace

• Abstractive approach

• Topic-based

• Different input data

Source code