Changes in version 0.1.1 (2026-08-20) Changes made in response to the CRAN review of 0.1.0. - themescope_cache_dir() no longer creates anything: it reports where models would be cached, and gained a create argument (default FALSE). The cache directory is created only when a model is actually downloaded. - ts_download_model() and ts_model_path() take model_dir = NULL instead of defaulting to a path under the user's home filespace. Pass a directory of your own, for instance tempdir(), to keep a download inside the session. - New ts_clear_cache() deletes the downloaded models and the cache directory, so everything the package stores on disk can be removed again. - The first download reports where it is caching. - Examples no longer use \dontrun{}. Those that need an internet connection use \donttest{} and fail gracefully, the Shiny launcher is guarded by if (interactive()), and every example that writes does so in tempdir(). Changes in version 0.1.0 First release. - The ThemeScope pipeline as a chain of exported functions: read_collection(), preprocess_texts(), build_vocab(), build_cooccurrence_matrix(), build_cooccurrence_network(), detect_communities(), compute_psi(), compute_cs(), term_relevance(), top_terms(), plot_themescope() and plot_network(), with themescope() running them end to end. - save_themescope() and read_themescope() write and reopen a .themescope archive holding the analysis, the annotated corpus, the raw texts, the term-level scores and the parameters of the run. - read_collection() imports csv, tsv, txt, xlsx, xls, RData and rda files, or a zip of many, and can renumber documents doc_1, doc_2, ... with sequential_ids = TRUE. - Updated Universal Dependencies 2.15 udpipe models are downloaded on demand and cached: ts_list_models(), ts_download_model(), ts_model_path(), themescope_cache_dir(). - themescope_app() launches an optional Shiny interface that calls the same exported functions, with a demo corpus that ships already annotated. - Bundled data: the brysbaert concreteness norms, a 1000-document Reddit climate sample and its udpipe annotation.