This paper describes about the our proposed system in the Consumer Health Information Search (CHIS) task. The objective of the task 1 is to classify the sentences in the document into relevant or irrelevant with respect to the query and task 2 is analysing the sentiment of the sentences in the documents with respect to the given query. In this proposed approach distributional representation of text along with its statistical and distance measures are carried over to perform the given tasks as a text classi cation problem. In our experiment, Non - Negative Matrix Factorization utilized to get the distributed representation of the document as well as queries, distance and correlation measures taken as the features and Random Forest Tree utilized to perform the classi cation. The proposed approach yields 70.19% in task 1 and 34.64% in task 2 as an average accuracy.
Over the past few years, tremendous amount of
investment and research carried on to enhance the predictive
analytics through text analytics in health care domain [
The informations explicitly available in Electronics Health Records (EHR) but implicitly available in clinical trails as a form of text. Now, our primary problem is becomes, representing text that can be easily and e ectively used for further 1https://medlineplus.gov/ 2https://clinicaltrials.gov/ 3https://clinicaltrials.gov/ application. The application may be a sequential modeling tasks (Information Extraction) or text classi cation tasks (Document Retrieval, sentiment analysis on retrieved documents and Validation of retrieved documents).
Document retrieval is primary task in text analytics application in which the Consumer Health Information Search (CHIS) is focused on validating the retrieved results (Relevant or Irrelevant) and performing sentiment analysis on retrieved results (Support, Oppose and Neutral). The given problem can be viewed as a text classi cation problem with the target classes as mentioned in above two tasks.
Text classi cation is a classic application in text analytics
domain, that is utilized in the multiple domains and
industries in various forms. Given a text content, the classi er
must have the capability of classifying it into the
predened set of classes [
The fundamental component in classi cation task is text
representation, which tries to represent the given text into its
equivalent form of numerical components. Later, these
numerical components are utilized directly for the classi cation
or will be used to extract the features required to perform the
classi cation task. This text representation methods evolved
over the time to improve the originality of representation,
which paves way to move from the frequency based
representation methods to the semantic representation methods.
Though other methods are also available, this paper focuses
only on Vector Space Model (VSM) and Vector Space Model
of Semantics (VSMs) [
In VSM, the text is represented as a vector, based on the
occurrence of terms (binary matrix) or frequency of the
occurrence of terms (Term - Document Matrix) present in the
given text. The given text is represented as a vector, based
on frequency of terms that occur within the text by having
vocabulary built across the entire corpus. Here, 'terms'
represents the words or the phrases [
The term documents matrix is ine cient due to the
biasing problem (i.e. few terms gets higher weight because of
unbalanced and uninformative data). To overcome this, Term
Frequency - Inverse Document Frequency (TF-IDF)
representation method is introduced, which re-weighs the term
frequency based upon its presence across the documents [
The Vector Space Model of Semantics (VSMs) overcomes
the above mentioned shortcomings by weighing terms based
on the context. This is achieved by applying TDM on
matrix factorization methods like Singular Value
Decomposition (SVD) and Non - Negative Matrix Factorization (NMF)
[
Word Embeddings along with the structure of the
sentence are utilized to represent the short texts. This requires
very less data and the dimension of the vector can be
controlled. To develop the Word to Vector (Word2Vec) model
it requires a very large corpus [
Considering above said pros and cons, here the proposed approach is experimented to observe the performance of distributional semantic representation of text in the classi cation task. The given query and documents are represented as a TDM matrix after the necessary preprocessing steps and NMF is applied on it to get the distributional representation. Thereafter, distance measure and correlation measures between query vector of each document and vector representation of the sentences in the documents are computed in order to perform the classi cation task.
This section describes about the distributional representation of the text, which is used further for the classi cation task. The distributional representation aims to compute the basis vector from the term frequency vector by applying NMF on the TDM. The systematic approach for the distributional representation is given in Figure 1. 2.1
Let, dk = s1; s2; s3; :::; sn are the sentences in the k th document in the document set D = d1; d2; d3; :::dn, qi represents the ith query and C = c1; c2; :::; cn are the classes in which s falls under with respect to the q and n is the size of corpus. The objective of the experimentation is to classify each sentence in the document into its respective prede ned classes. 2.2
Few of the terms that appears across multiple classes will shows con ict towards the classi cation, where the terms generally gets low weighs in TF-IDF representation. Hence these terms are eliminated if it occurs more than 3=4 times (1) (2) (3) across the classes and in order to avoid the sparsity of the representation, terms with the document frequency of one are eliminated. Here TF-IDF representation not considered. Because, it has a tendency to provide weighs for the rare words which is more common in clinical texts (Drug names, Measurements and Dosage levels). Here, advantage of the TF-IDF representation is indirectly obtained by handling document frequency of the terms. 2.3
In TDM, vocabulary has been computed by nding unique words present in the given corpus. Then the number of times term presents (term frequency) in each question is computed against the vocabulary formed. The terms present in this vocabulary acts as a rst level features.
A i;j = T DM (Corpus)
A i = termf requency(qi)
Where, i represents the ith sentence and j represents the jth term in the vocabulary. In-order to improve the representation, along with the unigram words, the bi-gram and tri-gram phrases also considered after following above mentioned preprocessing steps. 2.4
The above computed TDM is applied on NMF to get the distributional representation of the given corpus.
W i;r = nmf ( A i;j)
In general matrix factorization is done to get the product
of matrices, subject to their reconstruction that the error
needs to be low. The product components from the
factorization gives the characteristics of the original matrix [
A
W HT
If A is m n original TDM matrix, then W is i r basis matrix and H is j r mixture matrix. Linear combination of basis vectors (column vectors) of W along with the weights of H gives the approximated original matrix A. While factorizing, initially random values are assigned to W and H then the optimization function is applied on it to compute appropriate W and H.
minfr(W; H)
V
W HT 2
F s:t: W; H 0
Here F is Forbenius norm and r is parameter for
dimension reduction, which is set to be 10 to have i 10 xed size
vector for each question. Here NMF is used for nding out
the basis vector for the following reasons: the non-negativity
constraints makes interpretability straight forward than the
other factorization methods; selection of r is straight
forward; and the basis vector in semantic space is not
constrained to be orthogonal, which is not a ordable by nding
singular vectors or eigen vectors [
For this experiment the data set has been provided by
Consumer Health Information Search (CHIS) task
commit(4)
(5)
tee [
Task 1 : This task is becoming necessary unit in-order to lter the retrieved results from Information Retrieval (IR) application. This ensures the recall of the Search Engine which is mandatory in health care domain text analytics applications. With this information the remaining part of the section describes about the proposed approach in text classi cation in task 1.
Let, dk = s1; s2; s3; :::; sn are the sentences in the k th document in the document set as mentioned in the Table 1 (D = skincare; M M r; HRT; Ecig; V itc), qi represents the ith query and C = Relevant; Irrelevant are the classes which the s falls under with respect to the q. n is size of corpus and this is also mentioned in Table 1.
The objective of task is to classify the given question into its corresponding classes (Relevant, Irrelevant). The distributional representation of the given training and testing corpus are computed as described in the previous section. The systematic diagram for the remaining approach is given in Figure 2. After the representation, the similarity measures between query vector qi and sentence vectors in D are computed. The computed similarity measures are given in table 3. These similarity measures that is computed are taken as the attributes for the supervised classi cation algorithm which is Random Forest Tree (RFT).
By having typical f Cpf number of trees, output labels Y = y1; y2; y3; :::; yn (Relevant, Irrelevant) and feature set F = f 1; f 2; f 3; :::; f n the bagging repeatedly (B times Number of trees) done by selecting random samples and attributes from the training set and builds the decision tree for each set. Then the predictions for test set can be nd by averaging the predictions from all the individual decision trees built through the train set. It can be interpreted as following: fb = f (Wb; Yb; Fb)
B Y = 1 X fb(W^ F^)
B b=1 (6) (7)
In order to ensure the performance, 10-fold 10-cross validation performed during the training and this yields near 72% as a precision and it yields 68.12% against the test set.
Task 2 : This task is also necessary unit, in-order interpret further information from the retrieved results. This is task is similar to the task 1 and carried on exactly similar to the task 1 with target class labels as C = Oppose; Support; N eutral . The classes in C are the nal output label which the s falls under with respect to the q.
Here also 10-fold 10-cross validation performed during the training and this yields near 45% as a precision and it yields 38.53% against the test set. The detailed description about the results are given in Table 2.
The objective of the tasks (Consumer Health Information Search) are performed as a text classi cation problem based on the distributional representation of the text by utilizing # testing Sentences 90 60 74 66 74
qPid=1 jPi Qij2
miin jPi term - document matrix and non-negative matrix factorization. Even though the proposed approach not yields the state of art performance in the tasks, the obtained results are good enough to continue the research. These results are described in the Table 2. Distributional semantic representation methods su ers from the well known problem 'Curse of Dimensionality'. Hence the future work will be focused on reducing the dimensionality of the representation basis vectors and including the dedicated feature engineering for health care domain.