6.2 Beta 4: Bug in context-menu Selection tab

I came across this typo in The JPS Bible Commentary: Ecclesiastes (reported separately).

I right clicked on it to get to the Report Typo command, and noticed that the "Selection" tab included the open parenthesis as part of my selection even thought it wasn't selected. This is the bug I'm reporting here.

Find more posts tagged with

Comments

MJ. Smith

This is a result of the way the parser finds "phrases" the first "selection" is what you have highlighted; the second selection is the parser's judgment as to "phrase" which is less than ideal for scripture study due to its handling of punctuation. In short, think "by design". I'm not sure that we've been told which standard routine was implemented.

Rosie Perera

https://community.logos.com/discussion/comment/711553#Comment_711553

This is a result of the way the parser finds "phrases" the first "selection" is what you have highlighted; the second selection is the parser's judgment as to "phrase" which is less than ideal for scripture study due to its handling of punctuation. In short, think "by design". I'm not sure that we've been told which standard routine was implemented.

Well, if the typo weren't there, then "a wife" should be the phrase. With the typo, I'm not sure, but it's sort of a moot point I guess in this particular instance, because I hope they fix the typo. However it does point to an oddness in the underlying algorithm that it would pick up an opening parenthesis as the conclusion of a "phrase". If it's "by design" then I might come back and protest that the design should change.

Rosie Perera

https://community.logos.com/discussion/comment/711553#Comment_711553

This is a result of the way the parser finds "phrases"

I'm just trying to explore this parser some more and finding it doesn't seem to work at all in some books. I'm not sure if it's meant to work in all books or not. Here's what I'm seeing in some books. Everywhere in this particular book (Apostle Paul: His Life and Theology), no matter where I right click, the "phrase" option for SELECTION says "Paragraph". It can't seem to parse the phrases in this book for some reason. Should I report this as a bug, or is this "by design" too?

That is also the case with the book I have open in an adjacent tab, Pastor, Church & Law: Liability & Church and State Issues, Volume Four. But not the tab to the far left (Captive to the Word of God). I've been checking around randomly and it appears that about half the time it doesn't work. I haven't found any rhyme nor reason to when it doesn't work (e.g., these books it doesn't work in are not Vyrso eBooks).

EDIT: Bradley Grainger, is this a bug? I can start up a new thread for it if so.

MJ. Smith

https://community.logos.com/discussion/comment/711561#Comment_711561

However it does point to an oddness in the underlying algorithm that it would pick up an opening parenthesis as the conclusion of a "phrase". If it's "by design" then I might come back and protest that the design should change.

I'll simply say that Bradley and I agreed to disagree on whether this is a bug/by design that needs correction. My personal feeling is that the corpus for which the algorithm was designed/tested/trained is an inappropriate corpus to model the Logos corpus. But then again, I haven't seen the results of other common algorithms on the Logos corpus to evaluate the alternatives.

MJ. Smith

https://community.logos.com/discussion/comment/711566#Comment_711566

no matter where I right click, the "phrase" option for SELECTION says "Paragraph". It can't seem to parse the phrases in this book for some reason. Should I report this as a bug, or is this "by design" too?

I don't know the answer here. I don't understand what tagging was done just before the release of L6 that supported smart tags and this kind of phrase identification. I do not that it does not recognize clauses ... it is word, phrase, sentence, paragraph essentially. Certainly reporting it might get Bradley to give us more details so we know when reporting errors is appropriate.

Chris Culy (Logos)

https://community.logos.com/discussion/comment/711572#Comment_711572

The phrase data is built into the book itself, which requires a rebuild for all books released before smart selection came out. We have rebuilt many books, but no where near all of them. We know paragraph boundaries without doing this rebuild, so that's why that option is displaying in the context menu; it's likely that the book you have simply hasn't been rebuilt yet (that's my guess).

Bradley Grainger (Logos)

https://community.logos.com/discussion/comment/711571#Comment_711571

I'll simply say that Bradley and I agreed to disagree on whether this is a bug/by design that needs correction.

I don't think I've disagreed with you that smart text selection can have poor text boundaries.

Mike Binks

https://community.logos.com/discussion/comment/711687#Comment_711687

I'll simply say that Bradley and I agreed to disagree on whether this is a bug/by design that needs correction.

I don't think I've disagreed with you that smart text selection can have poor text boundaries.

Which, I ponder rhetorically, would be easier; fixing 'smart text selection' or editing the definition of 'smart' in each of the dictionaries that Logos carries?

MJ. Smith

https://community.logos.com/discussion/comment/711687#Comment_711687

I'll simply say that Bradley and I agreed to disagree on whether this is a bug/by design that needs correction.

I don't think I've disagreed with you that smart text selection can have poor text boundaries.

That is true and I do understand why it is unlikely to be changed in the near future but it is annoying when the most common error should have been picked up before you even got the algorithm - using an opening paired punctuation as a final element. But I know very well I couldn't even come close to building an acceptable algorithm so I gotta love those NLP nerds.

MJ. Smith

https://community.logos.com/discussion/comment/711660#Comment_711660

Thanks Chris, now I'll know how to answer this type of question.

Rosie Perera

https://community.logos.com/discussion/comment/711660#Comment_711660

Thanks, Chris.