You might want to consider the start of this tutorial.

Short introductions to other TF datasets:

• Dead Sea Scrolls,
• Old Babylonian Letters,

or the

• Quran

Search from MQL¶

Maybe you know MQL. It is the search template language implemented by Ulrik Sandborg-Petersen in Emdros, and used by SHEBANQ.

TF search templates have been inspired by MQL, but they are different.

This notebook shows examples of real-life MQL queries on SHEBANQ. and expresses them as Text-Fabric search templates.

See also the SHEBANQ tutorial by Bas Meeuse translated to TF.

We want to use a fixed data version, not a changing one, for the purposes of this notebook. Also, we pick it from our cloned repo. So, if you want to run this, you need to have cloned the bhsa repo to your computer.

By Oliver Glanz¶

MQL¶

Oliver Glanz: PP with adjective followed by noun

select all objects where
[phrase FOCUS typ = PP
  [word sp= prep]
  [word sp=adjv]
  [word sp=subs]
]

64 results having 251 words.

TF¶

Comparison¶

The number of results is right. The number of words that SHEBANQ reports is the number of words in the phrases of the result. Let us count them:

Inspired by Oliver Glanz¶

MQL¶

Dirk Roorda: Anapher

select all objects where
[phrase
  [word AS samelex FOCUS FIRST]
  [word FOCUS lex = samelex.lex]
]

220 results having 458 words.

This means: 220 verses, with 458 result words, and since there are exactly 2 words in each result, we expect 229 results.

TF¶

We miss 23 results. How can that be?

Well, in MQL

[phrase]
  [word]
  [word]

means that the two words are adjacent in the phrase. If the phrase has gaps, words around a gap are still adjacent.

In TF it is different, there such words are not adjacent.

Let's lift the requirement of adjacency from the TF query:

Way to many results.

Let's filter these results and only retain those where w1 and w2 are adjacent in the MQL sense.

So it looks like we have the same set in our hands.

There must be 23 cases of phrases with an intervening phrase after the first word, after which the phrase resumes with a word of the same lexeme. Let's identify those.

From search Gaps we pick the query that finds all gapped phrases, and we store it in a set.

Now we modify our loose query to run over gapped phrases only:

And apply the filter again:

Now we have one more than expected. It could be that one of the gapped phrases has an adjacent pair that does not lie around a gap.

Indeed, result 9 is such a case.

An other example by Oliver Glanz¶

We look for words with the same value for the g_cons feature but with a different lex feature. There is a word pair in Jeremiah 1:11-12 that has this.

An obvious query to try is the following:

But the thing is, one of those words is in verse 11, and the other in verse 12. The query as it stands, requires both words to be in the same verse.

We could write a bit more intricate query, that requires the word pair between the last word of verse 10 and the first one of verse 13, but this query turns out to be inefficient. It finds a result in a few seconds (that is already way to slow) and then it spends ages to search a vast space where no results are to be found.

An alternative is to look in a whole chapter, and to require the words not to be too far apart, e.g. at most 20.

Note that <20: relaxes to adjacency condition <: with 20 slots in both directions. If there is a pair of words within a slot distance of 20, then the reversed pair also satisfies that condition.

Hence we put in an extra w1 < w2, so that every pair occurs only once.

We can do so for the whole book of Jeremiah.

Let's try the whole Bible in this way:

TF is struggling with it, but it delivers!

By Martijn Naaijer¶

MQL¶

Martijn Naaijer: Object clauses with >CR

Select all objects where

[clause rela = Objc
   [word focus first lex = '>CR']
]

157 results

TF¶

Comparison¶

We have fewer cases: 96 instead of 157. We are working on the ETCBC version 2017, and the query has been executed against 4b. There have been coding updates that are relevant to this query, e.g. in Genesis 43:27, which is in the results on SHEBANQ, but not here. In 2017 the rela is Attr, and not Objc:

By Cody Kingham¶

MQL¶

Cody Kingham: Mutual Information Hierarchies. p.18n49. First Person Verbs in Narrative

SELECT ALL OBJECTS WHERE

[book
   [clause txt = 'N'
      [word FOCUS sp = verb
        [word ps = p1
         ]
      ]
   ]
]
OR
[book
   [clause txt = '?N'
      [word FOCUS sp = verb
        [word ps = p1
         ]
      ]
   ]
]

273 results.

TF¶

Comparison¶

One result less. Again, a coding difference between versions. Exercise: find out where that happened.

By Reinoud Oosting¶

MQL¶

Reinoud Oosting: to go + object marker

Select all objects
where
 [clause
   [phrase function = Pred OR function = PreC
     [word FOCUS sp = verb AND vs = qal AND lex = "HLK[" ]
         ]
    ..
    [phrase FOCUS
    [word First lex = ">T"]
   ]
 ]
OR
 [clause
    [phrase FOCUS
      [word First lex = ">T" ]
    ]
..
   [phrase function = Pred OR function = PreC
     [word FOCUS sp = verb AND vs = qal AND lex = "HLK["]
   ]
 ]

4 results.

TF¶

This is a case where we can simplify greatly because we are not hampered by automatic constraints on the order of the phrases.

By Reinoud Oosting (ii)¶

MQL¶

Reinoud Oosting: To establish covenant

select all objects
where

 [clause
   [phrase function = Pred OR function = PreC
     [word FOCUS sp = verb AND vs = hif AND lex = "QWM[" ]
         ]
    ..
    [phrase function = Objc
    [word FOCUS lex = "BRJT/" ]
   ]
 ]
OR
 [clause
    [phrase function = Objc
      [word FOCUS lex = "BRJT/" ]
    ]
..
   [phrase function = Pred OR function = PreC
     [word FOCUS sp = verb AND vs = hif AND lex = "QWM["]
   ]

]

13 results

TF¶

By Reinoud Oosting (iii)¶

MQL¶

Reinoud Oosting: To find grace in sight of

select all objects
where

 [clause
   [phrase FOCUS function = Pred OR function = PreC
     [word sp = verb AND vs = qal AND lex = "MY>[" ]
         ]
    ..
    [phrase function = Objc
    [word FOCUS lex = "XN/" ]
   ]
[phrase function = Cmpl
[word FOCUS lex = "B"]
[word FOCUS lex = "<JN/"]
]
 ]
OR
 [clause
    [phrase function = Objc
      [word FOCUS lex = "XN/" ]
    ]
[phrase function = Cmpl
[word FOCUS lex = "B"]
[word FOCUS lex = "<JN/"]
..
   [phrase function = Pred OR function = PreC
     [word FOCUS sp = verb AND vs = qal AND lex = "MY>["]
   ]
 ]
]

38 results

TF¶