Termvector

Mtas can produce termvectors for the set of documents satisfying the condition and/or filter. To get this information, in Solr requests, besides the parameter to enable the Mtas query component, the following parameter should be provided.

Parameter Value Obligatory
mtas.termvector true yes

Multiple termvector results can be produced within the same request. To distinguish them, a unique identifier has to be provided for each of the required document results.

Parameter Value Info Obligatory
mtas.termvector.<identifier>.key <string> key used in response no
mtas.termvector.<identifier>.field <string> Mtas field yes
mtas.termvector.<identifier>.prefix <string> prefix yes
mtas.termvector.<identifier>.number <double> number of terms in list no
mtas.termvector.<identifier>.start <string> begin list after provided term, only if sorted on term no
mtas.termvector.<identifier>.type <string> required type of statistics no
mtas.termvector.<identifier>.regexp <string> regular expression condition on term no
mtas.termvector.<identifier>.ignoreRegexp <string> regular expression condition for terms that have to be ignored no
mtas.termvector.<identifier>.sort.type <string> sort on term or type of statistics no
mtas.termvector.<identifier>.sort.direction <string> sort direction: asc or desc no

Full

When using distributed search, instead of applying the more efficient default algorithm where in two rounds lists of terms are collected and combined from the participating cores, also another approach can be used. Using the full option, the complete lists of terms (matching all requirements) is collected from the participating cores, and combined afterwards. This approach is likely to be less efficient when huge lists are involved, but necessary for example when results have to be sorted on specific statistics.

Parameter Value Info Obligatory
mtas.termvector.<identifier>.full <boolean> compute full list of terms no

List

If a list of terms is provided, the termvector will be restricted to items from this list. These items may be configured to be interpreted as explicit terms or as regular expressions.

Parameter Value Info Obligatory
mtas.termvector.<identifier>.list [<string>,…] list of terms yes
mtas.termvector.<identifier>.listRegexp <boolean> interpret items in provided list as regular expressions no

Furthermore, a list of terms can be provided that should be ignored within the termvector. These items may also be configured to be interpreted as explicit terms or as regular expressions.

Parameter Value Info Obligatory
mtas.termvector.<identifier>.ignoreList [<string>,…] list of terms yes
mtas.termvector.<identifier>.ignoreListRegexp <boolean> interpret items in provided ignoreList as regular expressions no

Distances

For each term in the termvector, the distance to a predefined base term can be computed.

Two type of distance are available: Levenshtein and Damerau–Levenshtein, each with configurable parameters to define the weight of the relevant operations.

Parameter Value Info Obligatory
mtas.termvector.<identifier>.distance.<identifier distance>.key <string> key used in response no
mtas.termvector.<identifier>.distance.<identifier distance>.type <string> type of distance yes
mtas.termvector.<identifier>.distance.<identifier distance>.base <string> base term for distance yes
mtas.termvector.<identifier>.distance.<identifier distance>.maximum <double> restrict termvector to terms with provided maximum no
mtas.termvector.<identifier>.distance.<identifier distance>.parameter.* <string> type dependent parameters no

The available type dependent additional parameters for type levenshtein and damerau-levenshtein are

Type Type dependent parameter Value Info Default
levenshtein, damerau-levenshtein deletionDistance <double> distance for a deletion 1.0
levenshtein, damerau-levenshtein insertionDistance <double> distance for an insertion 1.0
levenshtein, damerau-levenshtein replaceDistance <double> distance for a replacement 1.0
damerau-levenshtein transpositionDistance <double> distance for a transposition 1.0

Functions

Besides the specified statistics on hits over the documents, also statistics on the computed value of functions on the number of hits and the total number of words over the documents can be provided. In the definition of such a function, the number of hits is referred to as $q0, and the number of words is referred to as $n.

Parameter Value Info Obligatory
mtas.termvector.<identifier>.function.<identifier function>.key <string> key used in response no
mtas.termvector.<identifier>.function.<identifier function>.expression <string> definition of function yes
mtas.termvector.<identifier>.function.<identifier function>.type <string> required type of statistics no

Again, the key is added to the response and may be used to distinguish between multiple functions, and should therefore be unique within each specified termvector.


Examples

  1. Basic : basic statistics on occurring part of speech
  2. Regexp : words of length 5 containing only characters a-z, sorted descending by number of hits
  3. Ignore : previous result, ignoring words ending with $-e$.
  4. List : termvector for provided list of words.
  5. Start : termvector for words containing only characters a-z sorted by term and > koe.
  6. Distances : termvector for words with Levenshtein distance from 1 or less sorted descending by frequency.
  7. Functions : statistics on hits, relative frequency and total number of words in document for words containing only characters a-z.

Basic

Example
Total and average number of occurrences of part of speech (pos).

Request and response
q=*%3A*&mtas=true&mtas.termvector=true&mtas.termvector.0.key=example - basic&mtas.termvector.0.field=text&mtas.termvector.0.prefix=pos&mtas.termvector.0.number=3&rows=0&wt=json&indent=true

"mtas":{
    "termvector":[{
        "key":"example - basic",
        "list":[{
            "mean":200.22966889678833,
            "sum":25797991,
            "n":128842,
            "key":"ADJ"},
          {
            "mean":149.53835013602176,
            "sum":18689303,
            "n":124980,
            "key":"BW"},
          {
            "mean":459.93552395416265,
            "sum":59963634,
            "n":130374,
            "key":"LET"}]}]}

Regexp

Example
List of words with length 5 and containing only characters a-z, sorted descending by number of hits.

Regular expression
[a-z]{5}

Request and response
q=*%3A*&mtas=true&mtas.termvector=true&mtas.termvector.0.key=example - regexp&mtas.termvector.0.field=text&mtas.termvector.0.prefix=t_lc&mtas.termvector.0.number=5&mtas.termvector.0.type=n,sum&mtas.termvector.0.regexp=[a-z]{5}&mtas.termvector.0.sort.type=sum&rows=0&wt=json&indent=true

"mtas":{
    "termvector":[{
        "key":"example - regexp",
        "list":[{
            "sum":972687,
            "n":94160,
            "key":"heeft"},
          {
            "sum":645227,
            "n":84306,
            "key":"wordt"},
          {
            "sum":436038,
            "n":82453,
            "key":"onder"},
          {
            "sum":391488,
            "n":40512,
            "key":"zijne"},
          {
            "sum":314539,
            "n":62316,
            "key":"welke"}]}]}

Ignore

Example
List of words with length 5 and containing only characters a-z, sorted descending by number of hits, ignoring all words ending with $-e$.

Regular expressions
[a-z]{5} .*e

Request and response
q=*%3A*&mtas=true&mtas.termvector=true&mtas.termvector.0.key=example - ignore&mtas.termvector.0.field=text&mtas.termvector.0.prefix=t_lc&mtas.termvector.0.number=5&mtas.termvector.0.type=n,sum&mtas.termvector.0.regexp=[a-z]{5}&mtas.termvector.0.ignoreRegexp=.*e&mtas.termvector.0.sort.type=sum&rows=0&wt=json&indent=true

"mtas":{
    "termvector":[{
        "key":"example - ignore",
        "list":[{
            "sum":972687,
            "n":94160,
            "key":"heeft"},
          {
            "sum":645227,
            "n":84306,
            "key":"wordt"},
          {
            "sum":436038,
            "n":82453,
            "key":"onder"},
          {
            "sum":304620,
            "n":60555,
            "key":"leven"},
          {
            "sum":297160,
            "n":58263,
            "key":"waren"}]}]}

List

Example
Termvector for provided list of words.

List koe,paard,schaap,geit,kip

Request and response
q=*%3A*&mtas=true&mtas.termvector=true&mtas.termvector.0.key=example - list&mtas.termvector.0.field=text&mtas.termvector.0.prefix=t_lc&mtas.termvector.0.list=koe,paard,schaap,geit,kip&mtas.termvector.0.type=n,sum&rows=0&wt=json&indent=true

"mtas":{
    "termvector":[{
        "key":"example - list",
        "list":[{
            "sum":1128,
            "n":683,
            "key":"geit"},
          {
            "sum":1410,
            "n":864,
            "key":"kip"},
          {
            "sum":4432,
            "n":2344,
            "key":"koe"},
          {
            "sum":15478,
            "n":7436,
            "key":"paard"},
          {
            "sum":2154,
            "n":1591,
            "key":"schaap"}]}]}

Start

Example
Termvector for words containing only characters a-z sorted by term and > koe.

Request and response
q=*%3A*&mtas=true&mtas.termvector=true&mtas.termvector.0.key=example - start&mtas.termvector.0.field=text&mtas.termvector.0.prefix=t_lc&mtas.termvector.0.regexp=[a-z]*&mtas.termvector.0.number=5&mtas.termvector.0.start=koe&rows=0&wt=json&indent=true

"mtas":{
    "termvector":[{
        "key":"example - start",
        "list":[{
            "mean":2.0,
            "sum":2,
            "n":1,
            "key":"koea"},
          {
            "mean":1.0,
            "sum":1,
            "n":1,
            "key":"koeaan"},
          {
            "mean":1.0,
            "sum":2,
            "n":2,
            "key":"koeachtig"},
          {
            "mean":1.5,
            "sum":3,
            "n":2,
            "key":"koeachtige"},
          {
            "mean":1.0,
            "sum":2,
            "n":2,
            "key":"koeala"}]}]}

Distances

Example
List of words, sorted descending by number of hits, with at most Levenshtein distance 1 from regering. For each word, the Levenshtein distance from regering is computed, and also the Damerau-Levensthein distance from regering with specific weights for deletion, insertion, replacement and transposition.

Request and response
q=*%3A*&mtas=true&mtas.termvector=true&mtas.termvector.0.field=text&mtas.termvector.0.prefix=t_lc&mtas.termvector.0.key=distance&mtas.termvector.0.distance.0.type=levenshtein&mtas.termvector.0.distance.0.base=regering&mtas.termvector.0.distance.0.key=Levenshtein&mtas.termvector.0.distance.0.maximum=1&mtas.termvector.0.distance.1.type=damerau-levenshtein&mtas.termvector.0.distance.1.base=regering&mtas.termvector.0.distance.1.key=Damerau-Levenshtein&mtas.termvector.0.distance.1.parameter.deletionDistance=0.81&mtas.termvector.0.distance.1.parameter.insertionDistance=0.82&mtas.termvector.0.distance.1.parameter.replaceDistance=0.83&mtas.termvector.0.distance.1.parameter.transpositionDistance=0.84&mtas.termvector.0.number=5&mtas.termvector.0.sort.type=sum&mtas.termvector.0.sort.direction=desc&mtas.termvector.0.full=true&rows=0&wt=json

"mtas":{
    "termvector":[{
        "key":"distance",
        "listTotal":49,
        "list":[{
            "distance":{
              "Levenshtein":0.0,
              "Damerau-Levenshtein":0.0},
            "mean":1.7536344857153994,
            "sum":134979,
            "n":76971,
            "key":"regering"},
          {
            "distance":{
              "Levenshtein":1.0,
              "Damerau-Levenshtein":0.83},
            "mean":1.6863562423749492,
            "sum":16587,
            "n":9836,
            "key":"regeling"},
          {
            "distance":{
              "Levenshtein":1.0,
              "Damerau-Levenshtein":0.81},
            "mean":1.0262390670553936,
            "sum":352,
            "n":343,
            "key":"regerings"},
          {
            "distance":{
              "Levenshtein":1.0,
              "Damerau-Levenshtein":0.83},
            "mean":1.4080459770114941,
            "sum":245,
            "n":174,
            "key":"legering"},
          {
            "distance":{
              "Levenshtein":1.0,
              "Damerau-Levenshtein":0.81},
            "mean":1.0,
            "sum":97,
            "n":97,
            "key":"regering."}]}]}

Functions

Example
List of words containing only characters a-z, sorted descending by number of hits, with statistics on hits, relative frequency and total number of words in document.

Regular expression
[a-z]*

Functions
$q0/$n $n

Request and response
q=*%3A*&mtas=true&mtas.termvector=true&mtas.termvector.0.key=example - list&mtas.termvector.0.field=text&mtas.termvector.0.prefix=t_lc&mtas.termvector.0.regexp=[a-z]*&mtas.termvector.0.sort.type=sum&mtas.termvector.0.type=n,sum&mtas.termvector.0.function.0.expression=%24q0%2F%24n&mtas.termvector.0.function.0.key=relative+frequency&mtas.termvector.0.function.0.type=n%2Cmean&mtas.termvector.0.function.1.expression=%24n&mtas.termvector.0.function.1.key=total+number+of+words&mtas.termvector.0.function.1.type=n%2Csum&mtas.termvector.0.number=3&rows=0&wt=json&indent=true

"mtas":{
    "termvector":[{
        "key":"example - list",
        "list":[{
            "sum":15975272,
            "n":127444,
            "functions":{
              "total number of words":{
                "sum":391924648,
                "n":127444},
              "relative frequency":{
                "mean":0.040967994034336694,
                "n":127444}},
            "key":"de"},
          {
            "sum":10565895,
            "n":126197,
            "functions":{
              "total number of words":{
                "sum":391190126,
                "n":126197},
              "relative frequency":{
                "mean":0.028072930308247233,
                "n":126197}},
            "key":"van"},
          {
            "sum":8798835,
            "n":125415,
            "functions":{
              "total number of words":{
                "sum":391306760,
                "n":125415},
              "relative frequency":{
                "mean":0.02376864203286862,
                "n":125415}},
            "key":"en"}]}]}

Lucene

To use termvectors directly in Lucene, ComponentTermvector together with the provided collect method can be used.