Bridging constrained decoding for embodied vision-language planning and computational models of language acquisition
2
papers here
2
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
18
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-16) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
No output score is available yet. The snapshot lists 69 works in the last three years but lacks a complete, venue-level output breakdown. This axis would measure total publication volume and its distribution across venues and output types.
Impact—
The impact axis is unscored. The snapshot carries an h-index of 18 and an i10-index of 33, but per-paper citation evidence is too thin for a reliable score. This axis would measure citation influence and uptake of this researcher's work.
Focus0/100
The focus score of 0 reflects an even 50/50 split between two fields in the snapshot: embodied AI with constrained decoding, and computational linguistics with language acquisition. Topic counts likewise spread across topic modeling (89), multimodal machine learning applications (78), NLP techniques (77), logic and reasoning (38), and categorization, perception, and language (26), indicating broad rather than narrowly concentrated work.
Collaboration28/100
This axis scores 28, based on a low distinct-coauthor count of 3 in the resolved snapshot. Work concentrates in a small, stable cluster with Tianyi Ma, Ellie Xia, and Alan Hezao Ke — the frequent coauthors on the reviewed papers. Such compact teams suit constrained-decoding and language-acquisition studies, but the dataset may undercount coauthors outside the OpenAlex-resolved record.
Activity100/100
The activity score of 100 reflects a steep recent upswing in the year-by-year counts: from 18 works in 2022 and 21 in 2024 to 32 in 2025, with 16 already recorded for 2026, and 69 works over the last three years. This is the highest sustained output level in the 2007–2026 window covered by the snapshot.
Reproducibility38/100
The reproducibility score of 38 is the mean over 2 Evolve-reviewed papers. The telicity-acquisition study rates low — its GPT-2 labeling pipeline raises circularity concerns noted in the review — while CLAMP rates medium, with decode-time constraints and HMM lookahead that are implementable despite mixed reported metrics. A two-paper base keeps confidence modest.
Research summary
Parisa Kordjamshidi is a researcher at Michigan State University (US) working across two halves of the language-and-reasoning spectrum. In the snapshot's field distribution, embodied AI with constrained decoding and computational linguistics with language acquisition each account for 50% of the resolved record. The Evolve catalog reviews two 2026 arXiv preprints: CLAMP, which adds decode-time syntax, visibility, and reachability constraints plus HMM lookahead to frozen vision-language planners (last author position, medium reproducibility readiness), and a study modeling the developmental shift in telicity acquisition using CHILDES, surprisal, and GPT-2 (middle position, low readiness). Snapshot indicators from the OpenAlex-resolved profile include an h-index of 18, an i10-index of 33, and 69 works in the last three years. Activity is the strongest radar dimension (100): annual work counts climb from 18 in 2022 and 21 in 2024 to 32 in 2025, with 16 already logged for 2026. Topic counts concentrate on topic modeling (89), multimodal machine learning applications (78), NLP techniques (77), logic, reasoning, and knowledge (38), and categorization, perception, and language (26). Collaboration scores lower (28) with 3 distinct coauthors, anchored by Tianyi Ma, Ellie Xia, and Alan Hezao Ke. Output and impact axes remain unscored pending fuller citation data.
Collaboration pattern
Within the resolved snapshot, collaboration is compact: 3 distinct coauthors, with Tianyi Ma, Ellie Xia, and Alan Hezao Ke as the recurring partners on the reviewed work. The collaboration radar score is 28, and both 2026 preprints list Kordjamshidi in non-corresponding positions (middle on the telicity study, last on CLAMP), consistent with senior roles in a small team rather than large consortium-style projects.
Research keywords
Highlights
Decode-time constraints with HMM lookahead for frozen VLM planners
Modeling the developmental shift in telicity acquisition
Sustained publication upswing since 2022
Annual work counts in the snapshot rise from 18 (2022) and 21 (2024) to 32 (2025), with 16 recorded for 2026 and 69 works over the last three years — the profile's strongest radar dimension (activity: 100).
Timeline
- 2007Earliest works appear in the snapshot's year-by-year counts, with 2 works recorded.
- 201511 works recorded — an early peak in the annual counts.
- 202218 works, marking the start of a steep upward climb.
- 202421 works recorded, continuing the acceleration.
- 202532 works — the highest annual count in the snapshot.
- 2026Two Evolve-reviewed arXiv preprints appear: CLAMP on constrained decoding for embodied planning, and a study modeling the developmental shift in telicity acquisition.
Papers in the Evolve catalog
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometrics come from an OpenAlex-resolved profile only; name-based resolution may miss or merge records.
- Radar output and impact axes are unscored due to insufficient underlying data; h-index (18) and i10-index (33) are snapshot-level indicators only.
- Reproducibility rests on just 2 Evolve-reviewed papers (mean 38), one of which is rated low readiness.
- The 2026 count (16 works) covers a partial year, and the reviewed preprints are arXiv postings not yet peer-reviewed.