Improve Sentence Delineation in Resources
Robert Kelbe
Member Posts: 585 ✭✭✭
When you click anywhere in a resource, the Mobile App automatically highlights the entire sentence. I love this behavior, as it allows me to quickly highlight sentences.
Often, however, Logos is not able to select the sentence correctly. In most older resources, the sentence selection includes the trailing space after the sentence, and I have to manually deselect the trailing space. (In newer resources, it works as expected). Additionally, complex formatting seems to confuse the sentence selection. For example, if a sentence ends with a quotation mark following the period, both sentences are selected as if they are a single sentence. On the other hand, with abbreviations like "e.g.", the sentence often stops at the first period in the abbreviation, thus dividing one sentence into two. Other times, a large chunk of text (dozens of sentences) is included in one "sentence", for reasons unknown.
Based on this behavior, I assume each resource comes with the precomputed sentence delineation information. Older resources computed the sentence delineation with an older algorithm that included the trailing space. The algorithm also must get confused with complex punctuation like quotation marks and common abbreviations.
Please consider improving the sentence delineation algorithm and rebuilding all resources to have consistently better sentence delineation across all resources.
See forum post here: https://community.logos.com/forums/t/209513.aspx?PageIndex=1
Often, however, Logos is not able to select the sentence correctly. In most older resources, the sentence selection includes the trailing space after the sentence, and I have to manually deselect the trailing space. (In newer resources, it works as expected). Additionally, complex formatting seems to confuse the sentence selection. For example, if a sentence ends with a quotation mark following the period, both sentences are selected as if they are a single sentence. On the other hand, with abbreviations like "e.g.", the sentence often stops at the first period in the abbreviation, thus dividing one sentence into two. Other times, a large chunk of text (dozens of sentences) is included in one "sentence", for reasons unknown.
Based on this behavior, I assume each resource comes with the precomputed sentence delineation information. Older resources computed the sentence delineation with an older algorithm that included the trailing space. The algorithm also must get confused with complex punctuation like quotation marks and common abbreviations.
Please consider improving the sentence delineation algorithm and rebuilding all resources to have consistently better sentence delineation across all resources.
See forum post here: https://community.logos.com/forums/t/209513.aspx?PageIndex=1
1