transform - Transform documents into lower-dimensional space - MATLAB (original) (raw)
Transform documents into lower-dimensional space
Syntax
Description
[dscores](#d126e59645) = transform([lsaMdl](#d126e59409),[documents](#d126e59446)) transforms documents into the semantic space of the latent semantic analysis (LSA) model lsaMdl.
[dscores](#d126e59645) = transform([lsaMdl](#d126e59409),[bag](#d126e59470)) transforms documents represented by the bag-of-words or bag-of-n-grams modelbag into the semantic space of the LSA modellsaMdl.
[dscores](#d126e59645) = transform([lsaMdl](#d126e59409),[counts](#d126e59504)) transforms documents represented by the matrix of word counts into the semantic space of the LSA model lsaMdl.
[dscores](#d126e59645) = transform([ldaMdl](#d126e59423),[documents](#d126e59446)) transforms documents into the latent Dirichlet allocation (LDA) topic probability space of LDA model ldaMdl. The rows ofdscores are the topic mixture representations of the documents.
[dscores](#d126e59645) = transform([ldaMdl](#d126e59423),[bag](#d126e59470)) transforms documents represented by the bag-of-words or bag-of-n-grams modelbag into the LDA topic probability space of LDA modelldaMdl.
[dscores](#d126e59645) = transform([ldaMdl](#d126e59423),[counts](#d126e59504)) transforms documents represented by the matrix of word counts into the LDA topic probability space of LDA model ldaMdl.
[dscores](#d126e59645) = transform(___,[Name,Value](#namevaluepairarguments)) specifies additional options using one or more name-value pair arguments. These name-value pairs only apply if the input model is an ldaModel object.
Examples
Load the example data. The file sonnetsPreprocessed.txt contains preprocessed versions of Shakespeare's sonnets. The file contains one sonnet per line, with words separated by a space. Extract the text from sonnetsPreprocessed.txt, split the text into documents at newline characters, and then tokenize the documents.
filename = "sonnetsPreprocessed.txt"; str = extractFileText(filename); textData = split(str,newline); documents = tokenizedDocument(textData);
Create a bag-of-words model using bagOfWords.
bag = bagOfWords(documents)
bag = bagOfWords with properties:
NumWords: 3092
Counts: [154×3092 double]
Vocabulary: ["fairest" "creatures" "desire" "increase" "thereby" "beautys" "rose" "might" "never" "die" "riper" "time" "decease" "tender" "heir" "bear" "memory" "thou" … ] (1×3092 string)
NumDocuments: 154Fit an LSA model with 20 components.
numCompnents = 20; mdl = fitlsa(bag,numCompnents)
mdl = lsaModel with properties:
NumComponents: 20
ComponentWeights: [2.7866e+03 515.5889 443.6428 316.4191 295.4065 261.8927 226.1649 186.2160 170.6413 156.6033 151.5275 146.2553 141.6741 135.5318 134.1694 128.9931 124.2382 122.2931 116.5035 116.2590]
DocumentScores: [154×20 double]
WordScores: [3092×20 double]
Vocabulary: ["fairest" "creatures" "desire" "increase" "thereby" "beautys" "rose" "might" "never" "die" "riper" "time" "decease" "tender" "heir" "bear" "memory" … ] (1×3092 string)
FeatureStrengthExponent: 2Use transform to transform the first 10 documents into the semantic space of the LSA model.
dscores = transform(mdl,documents(1:10))
dscores = 10×20
5.6059 -1.8559 0.9286 -0.7086 -0.4652 0.8340 0.6751 -0.0611 -0.2268 1.9320 -0.7289 -1.0864 0.7131 -0.0571 -0.3401 0.0940 -0.4406 1.7507 -1.1534 0.1785
7.3069 -2.3578 1.8359 -2.3442 -1.5776 2.0310 0.7948 -1.3411 1.1700 1.8839 0.0883 0.4734 -1.1244 0.6795 1.3585 -0.0247 0.3627 -0.5414 -0.0272 -0.0114
7.1056 -2.3508 -2.8837 -1.0688 -0.3462 0.6962 0.0334 0.0472 -0.4916 0.6496 -1.1959 -1.0171 -0.4020 1.2953 -0.4583 0.5984 -0.3890 1.1780 0.6413 0.6575
8.6292 -3.0471 -0.8512 -0.4356 -0.3055 -0.4671 -1.4219 0.8454 0.8270 0.4122 2.2082 -1.1770 1.7775 -2.2344 -2.7813 1.4979 0.7486 -2.0593 0.6376 1.0721
1.0434 1.7490 0.8703 -2.2315 -1.1221 -0.2848 -2.0522 0.6975 -1.7191 -0.2852 0.8879 0.9950 -0.5555 0.8842 -0.0360 1.0050 0.4158 0.5061 0.9602 0.4672
6.8358 -2.0806 -3.3798 -1.0452 -0.2075 -2.0970 -0.4477 -0.2080 -0.9532 1.6203 0.6653 0.0036 1.0825 0.6396 -0.2154 -0.0794 0.7108 1.8007 -4.0326 -0.3872
2.3847 0.3923 -0.4323 -1.5340 0.4023 1.0396 -1.0326 -0.3776 -0.2101 -1.0944 -0.7513 -0.2894 0.4303 0.1864 0.4922 0.4844 0.5191 -0.2378 0.9528 0.4817
3.7925 -0.3941 -4.4610 -0.4930 0.4651 -0.3404 -0.5493 -0.1470 -0.5065 0.2566 0.3394 -1.1529 -0.0391 -0.8800 -0.4712 0.9672 0.5457 -0.3639 -0.3085 0.5637
4.6522 0.7188 -1.1787 -0.8996 0.3360 -0.4531 -0.1935 -0.3328 0.8640 -1.6679 -0.8056 -2.1993 0.1808 0.0163 -0.9520 -0.8982 0.6603 3.6451 1.2412 1.9621
8.8218 -0.8168 -2.5101 1.1197 -0.8673 1.2336 0.0768 -0.1943 0.7629 -0.1222 0.3786 1.1611 0.2326 0.3415 -0.3327 -0.3792 1.7554 0.2526 -2.1574 -0.0193To reproduce the results in this example, set rng to 'default'.
Load the example data. The file sonnetsPreprocessed.txt contains preprocessed versions of Shakespeare's sonnets. The file contains one sonnet per line, with words separated by a space. Extract the text from sonnetsPreprocessed.txt, split the text into documents at newline characters, and then tokenize the documents.
filename = "sonnetsPreprocessed.txt"; str = extractFileText(filename); textData = split(str,newline); documents = tokenizedDocument(textData);
Create a bag-of-words model using bagOfWords.
bag = bagOfWords(documents)
bag = bagOfWords with properties:
NumWords: 3092
Counts: [154×3092 double]
Vocabulary: ["fairest" "creatures" "desire" "increase" "thereby" "beautys" "rose" "might" "never" "die" "riper" "time" "decease" "tender" "heir" "bear" "memory" "thou" … ] (1×3092 string)
NumDocuments: 154Fit an LDA model with five topics.
numTopics = 5; mdl = fitlda(bag,numTopics)
Initial topic assignments sampled in 0.035174 seconds.
| Iteration | Time per | Relative | Training | Topic | Topic | | | iteration | change in | perplexity | concentration | concentration | | | (seconds) | log(L) | | | iterations |
| 0 | 0.00 | | 1.212e+03 | 1.250 | 0 | | 1 | 0.01 | 1.2300e-02 | 1.112e+03 | 1.250 | 0 | | 2 | 0.01 | 1.3254e-03 | 1.102e+03 | 1.250 | 0 | | 3 | 0.01 | 2.9402e-05 | 1.102e+03 | 1.250 | 0 |
mdl = ldaModel with properties:
NumTopics: 5
WordConcentration: 1
TopicConcentration: 1.2500
CorpusTopicProbabilities: [0.2000 0.2000 0.2000 0.2000 0.2000]
DocumentTopicProbabilities: [154×5 double]
TopicWordProbabilities: [3092×5 double]
Vocabulary: ["fairest" "creatures" "desire" "increase" "thereby" "beautys" "rose" "might" "never" "die" "riper" "time" "decease" "tender" "heir" "bear" "memory" … ] (1×3092 string)
TopicOrder: 'initial-fit-probability'
FitInfo: [1×1 struct]Use transform to transform the documents into a vector of topic probabilities. You can visualize these mixtures using stacked bar charts. View the topic mixtures of the first 10 documents.
topicMixtures = transform(mdl,documents(1:10)); figure barh(topicMixtures,'stacked') xlim([0 1]) title("Topic Mixtures") xlabel("Topic Probability") ylabel("Document") legend("Topic " + string(1:numTopics),'Location','northeastoutside')
Load the example data. sonnetsCounts.mat contains a matrix of word counts and a corresponding vocabulary of preprocessed versions of Shakespeare's sonnets.
load sonnetsCounts.mat size(counts)
Fit an LDA model with 20 topics. To reproduce the results in this example, set rng to 'default'.
rng('default') numTopics = 20; mdl = fitlda(counts,numTopics)
Initial topic assignments sampled in 0.041646 seconds.
| Iteration | Time per | Relative | Training | Topic | Topic | | | iteration | change in | perplexity | concentration | concentration | | | (seconds) | log(L) | | | iterations |
| 0 | 0.00 | | 1.159e+03 | 5.000 | 0 | | 1 | 0.02 | 5.4884e-02 | 8.028e+02 | 5.000 | 0 | | 2 | 0.02 | 4.7400e-03 | 7.778e+02 | 5.000 | 0 | | 3 | 0.02 | 3.4597e-03 | 7.602e+02 | 5.000 | 0 | | 4 | 0.02 | 3.4662e-03 | 7.430e+02 | 5.000 | 0 | | 5 | 0.02 | 2.9259e-03 | 7.288e+02 | 5.000 | 0 | | 6 | 0.02 | 6.4180e-05 | 7.291e+02 | 5.000 | 0 |
mdl = ldaModel with properties:
NumTopics: 20
WordConcentration: 1
TopicConcentration: 5
CorpusTopicProbabilities: [0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500 0.0500]
DocumentTopicProbabilities: [154×20 double]
TopicWordProbabilities: [3092×20 double]
Vocabulary: ["1" "2" "3" "4" "5" "6" "7" "8" "9" "10" "11" "12" "13" "14" "15" "16" "17" "18" "19" "20" "21" "22" "23" "24" "25" "26" … ] (1×3092 string)
TopicOrder: 'initial-fit-probability'
FitInfo: [1×1 struct]Use transform to transform the documents into a vector of topic probabilities.
topicMixtures = transform(mdl,counts(1:10,:))
topicMixtures = 10×20
0.0167 0.0035 0.1645 0.0977 0.0433 0.0833 0.0987 0.0033 0.0299 0.0234 0.0033 0.0345 0.0235 0.0958 0.0667 0.0167 0.0300 0.0519 0.0833 0.0300
0.0711 0.0544 0.0116 0.0044 0.0033 0.0033 0.0431 0.0053 0.0145 0.0421 0.0971 0.0033 0.0040 0.1632 0.1784 0.0937 0.0683 0.0398 0.0954 0.0037
0.0293 0.0482 0.1078 0.0322 0.0036 0.0036 0.0464 0.0036 0.0064 0.0612 0.0036 0.0176 0.0036 0.0464 0.0906 0.1169 0.0888 0.1115 0.1180 0.0607
0.0055 0.0962 0.2403 0.0033 0.0296 0.1613 0.0164 0.0955 0.0163 0.0045 0.0172 0.0033 0.0415 0.0404 0.0342 0.0176 0.0417 0.0642 0.0033 0.0676
0.0341 0.0224 0.0341 0.0645 0.0948 0.0038 0.0189 0.1099 0.0187 0.0560 0.1045 0.0356 0.0668 0.1196 0.0038 0.0931 0.0493 0.0038 0.0038 0.0626
0.0445 0.0035 0.1167 0.0034 0.0446 0.0583 0.1268 0.0169 0.0034 0.1135 0.0034 0.0034 0.0047 0.0993 0.0909 0.0582 0.0308 0.0887 0.0856 0.0034
0.1720 0.0764 0.0090 0.0180 0.0325 0.1213 0.0036 0.0036 0.0505 0.0472 0.0348 0.0477 0.0039 0.0038 0.0122 0.0041 0.0036 0.1605 0.1487 0.0465
0.0043 0.0033 0.1248 0.0033 0.0299 0.0033 0.0690 0.1699 0.0695 0.0982 0.0033 0.0039 0.0620 0.0833 0.0040 0.0700 0.0033 0.1479 0.0033 0.0433
0.0412 0.0387 0.0555 0.0165 0.0166 0.0433 0.0033 0.0038 0.0048 0.0033 0.0473 0.0474 0.1290 0.1107 0.0089 0.0112 0.0167 0.1555 0.2423 0.0040
0.0362 0.0035 0.1117 0.0304 0.0034 0.1248 0.0439 0.0340 0.0168 0.0714 0.0034 0.0214 0.0056 0.0449 0.1438 0.0036 0.0290 0.1437 0.0980 0.0304Input Arguments
Input LSA model, specified as an lsaModel object.
Input LDA model, specified as an ldaModel object.
Input documents, specified as a tokenizedDocument array, a string array of words, or a cell array of character vectors. If documents is atokenizedDocument, then it must be a column vector. Ifdocuments is a string array or a cell array of character vectors, then it must be a row of the words of a single document.
Tip
To ensure that the function does not discard useful information, you must first preprocess the input documents using the same steps used to preprocess the documents used to train the model.
Input bag-of-words or bag-of-n-grams model, specified as a bagOfWords object or a bagOfNgrams object. If bag is abagOfNgrams object, then the function treats each n-gram as a single word.
Frequency counts of words, specified as a matrix of nonnegative integers. If you specify'DocumentsIn' to be 'rows', then the valuecounts(i,j) corresponds to the number of times the_j_th word of the vocabulary appears in the _i_th document. Otherwise, the value counts(i,j) corresponds to the number of times the _i_th word of the vocabulary appears in the_j_th document.
Name-Value Arguments
Specify optional pairs of arguments asName1=Value1,...,NameN=ValueN, where Name is the argument name and Value is the corresponding value. Name-value arguments must appear after other arguments, but the order of the pairs does not matter.
Before R2021a, use commas to separate each name and value, and enclose Name in quotes.
Example: 'IterationLimit',200 sets the iteration limit to 200.
Note
These name-value pairs only apply if the input model is anldaModel object.
Orientation of documents in the word count matrix, specified as the comma-separated pair consisting of 'DocumentsIn' and one of the following:
'rows'– Input is a matrix of word counts with rows corresponding to documents.'columns'– Input is a transposed matrix of word counts with columns corresponding to documents.
This option only applies if you specify the input documents as a matrix of word counts.
Note
If you orient your word count matrix so that documents correspond to columns and specify'DocumentsIn','columns', then you might experience a significant reduction in optimization-execution time.
Maximum number of iterations, specified as the comma-separated pair consisting of 'IterationLimit' and a positive integer.
Example: 'IterationLimit',200
Relative tolerance on log-likelihood, specified as the comma-separated pair consisting of 'LogLikelihoodTolerance' and a positive scalar. The optimization terminates when this tolerance is reached.
Example: 'LogLikelihoodTolerance',0.001
Output Arguments
Output document scores, returned as a matrix of score vectors.
Version History
Introduced in R2017b