Generating and ranking incremental search suggestions for personal content
View Patent ↗Providing incremental search suggestions from a content database include accessing the content database to determine possible candidates for the search suggestions, scoring each of the candidates based at least partially on a non-monotonic document frequency function, where candidates that appear a first amount corresponding to a relatively frequent occurrence in the content database and candidates that appear a second amount corresponding to a relatively infrequent occurrence in the content database both score lower than candidates that appear in the content database with a frequency that is between the first amount and the second amount, and ordering the possible candidates based on at least the scoring. Possible candidates may include named entities and n-grams. The n-grams may only be unigrams and bigrams. Stop words may be filtered out of the n-grams. Scoring may include taking into account term frequency.
1. A method of providing incremental search suggestions from a content database, comprising:
accessing the content database to determine possible candidates for the search suggestions;
scoring each of the candidates based at least partially on a non-monotonic document frequency function, wherein candidates that appear a first amount corresponding to a relatively frequent occurrence in the content database and candidates that appear a second amount corresponding to a relatively infrequent occurrence in the content database both score lower than candidates that appear in the content database with a frequency that is between the first amount and the second amount; and
ordering the possible candidates based on at least the scoring.
2. A method, according to claim 1 , wherein possible candidates include named entities and n-grams.
3. A method, according to claim 2 , wherein the n-grams are only unigrams and bigrams.
4. A method, according to claim 2 , wherein stop words are filtered out of the n-grams.
5. A method, according to claim 1 , wherein scoring includes taking into account term frequency.
6. A method, according to claim 5 , wherein a boost factor is applied to change scores associated with named entities.
7. A method, according to claim 5 , wherein a boost factor is applied to change scores depending upon which of a number of possible parts of a document contains a corresponding candidate.
8. A method, according to claim 7 , wherein the parts of a document include a heading, a tag, a body, a footnote, an endnote, a comment, and an attachment.
9. A method, according to claim 1 , wherein the content database is private or semi-private corporate or personal content database.
10. A method, according to claim 9 , wherein the content database is provided by one of: the Evernote content management software and service and the OneNote® note-taking software product.
11. A method, according to claim 1 , wherein the incremental search suggestions include partially typed search terms.
12. A method, according to claim 11 , wherein the partially typed search terms are at least one of: prefixes and arbitrary contiguous fragments that correspond to possible candidates.
13. A method, according to claim 1 , wherein the incremental search suggestions are expanded to include alternative terms driven by various types of semantic relevance.
14. A non-transitory computer-readable medium containing software that provides incremental search suggestions from a content database, the software comprising:
executable code that accesses the content database to determine possible candidates for the search suggestions;
executable code that scores each of the candidates based at least partially on a non-monotonic document frequency function, wherein candidates that appear a first amount corresponding to a relatively frequent occurrence in the content database and candidates that appear a second amount corresponding to a relatively infrequent occurrence in the content database both score lower than candidates that appear in the content database with a frequency that is between the first amount and the second amount; and
executable code that orders the possible candidates based on at least the scoring.
15. A non-transitory computer-readable medium, according to claim 14 , wherein possible candidates include named entities and n-grams.
16. A non-transitory computer-readable medium, according to claim 15 , wherein the n-grams are only unigrams and bigrams.
17. A non-transitory computer-readable medium, according to claim 15 , wherein stop words are filtered out of the n-grams.
18. A non-transitory computer-readable medium, according to claim 14 , wherein scoring includes taking into account term frequency.
19. A non-transitory computer-readable medium, according to claim 18 , wherein a boost factor is applied to change scores associated with named entities.
20. A non-transitory computer-readable medium, according to claim 18 , wherein a boost factor is applied to change scores depending upon which of a number of possible parts of a document contains a corresponding candidate.
21. A non-transitory computer-readable medium, according to claim 20 , wherein the parts of a document include a heading, a tag, a body, a footnote, an endnote, a comment, and an attachment.
22. A non-transitory computer-readable medium, according to claim 14 , wherein the content database is private or semi-private corporate or personal content database.
23. A non-transitory computer-readable medium, according to claim 22 , wherein the content database is provided by one of: the Evernote content management software and service and the OneNote® note-taking software product.
24. A non-transitory computer-readable medium, according to claim 14 , wherein the incremental search suggestions include partially typed search terms.
25. A non-transitory computer-readable medium, according to claim 24 , wherein the partially typed search terms are at least one of: prefixes and arbitrary contiguous fragments that correspond to possible candidates.
26. A non-transitory computer-readable medium, according to claim 14 , wherein the incremental search suggestions are expanded to include alternative terms driven by various types of semantic relevance.