Constructing a sentiment analysis model for LibQUAL+ comments
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Purpose The purpose of this paper is to establish a data mining model for performing sentiment analysis on open-ended qualitative LibQUAL+ comments, providing a further method for year-to-year comparison of user satisfaction, both of the library as a whole and individual topics. Design/methodology/approach A training set of 514 comments, selected at random from five LibQUAL+ survey responses, was manually reviewed and labeled as having a positive or negative sentiment. Using the open-source RapidMiner data mining platform, those comments provided the framework for creating library-specific positive and negative word vectors to power the sentiment analysis model. A further process was created to help isolate individual topics within the larger comments, allowing for more nuanced sentiment analysis. Findings Applied to LibQUAL+ comments for a Canadian mid-sized academic research library, the model suggested a fairly even distribution of positive and negative sentiment in overall comments. When filtering comments into affect of service, information control and library as place, the three dimensions’ relative polarity mirrored the results of the quantitative LibQUAL+ questions, with highest scores for affect of service and lowest for library as place. Practical implications The sentiment analysis model provides a complementary tool to the LibQUAL+ quantitative results, allowing for simple, time-efficient, year-to-year analysis of open-ended comments. Furthermore, the process provides the means to isolate specific topics based on specified keywords, allowing individual institutions to tailor results for more in-depth analysis. Originality/value To best account for library-specific terminology and phrasing, the sentiment model was created using LibQUAL+ open-ended comments as the foundation for the sentiment model’s classification process. The process also allows individual topics, chosen to meet individual library needs, to be isolated and independently analyzed, providing more precise examination.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it